CN118539616A - Power load remote monitoring and early warning method and system based on Internet of things - Google Patents
Power load remote monitoring and early warning method and system based on Internet of things Download PDFInfo
- Publication number
- CN118539616A CN118539616A CN202410980734.4A CN202410980734A CN118539616A CN 118539616 A CN118539616 A CN 118539616A CN 202410980734 A CN202410980734 A CN 202410980734A CN 118539616 A CN118539616 A CN 118539616A
- Authority
- CN
- China
- Prior art keywords
- power load
- abnormal
- data
- user
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000012544 monitoring process Methods 0.000 title claims abstract description 48
- 230000002159 abnormal effect Effects 0.000 claims abstract description 151
- 238000005070 sampling Methods 0.000 claims abstract description 103
- 230000005856 abnormality Effects 0.000 claims abstract description 68
- 238000012549 training Methods 0.000 claims abstract description 61
- 230000004044 response Effects 0.000 claims abstract description 22
- 230000005611 electricity Effects 0.000 claims abstract description 13
- 230000008859 change Effects 0.000 claims description 91
- 230000007774 longterm Effects 0.000 claims description 19
- 230000000694 effects Effects 0.000 claims description 18
- 238000004458 analytical method Methods 0.000 claims description 16
- 230000001186 cumulative effect Effects 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 6
- 239000003102 growth factor Substances 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000006855 networking Effects 0.000 claims 2
- 238000001514 detection method Methods 0.000 description 18
- 230000035945 sensitivity Effects 0.000 description 14
- 238000012545 processing Methods 0.000 description 12
- 238000012423 maintenance Methods 0.000 description 11
- 238000007726 management method Methods 0.000 description 10
- 238000005457 optimization Methods 0.000 description 10
- 238000013461 design Methods 0.000 description 9
- 238000009825 accumulation Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 6
- 230000007246 mechanism Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 230000004043 responsiveness Effects 0.000 description 5
- 238000013500 data storage Methods 0.000 description 4
- 230000007423 decrease Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 238000005265 energy consumption Methods 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000035772 mutation Effects 0.000 description 2
- 230000002688 persistence Effects 0.000 description 2
- 230000002085 persistent effect Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000000700 time series analysis Methods 0.000 description 1
Landscapes
- Remote Monitoring And Control Of Power-Distribution Networks (AREA)
- Telephonic Communication Services (AREA)
Abstract
The invention provides a power load remote monitoring and early warning method and system based on the Internet of things, and relates to the technical field of power load early warning, wherein the method comprises the following steps: setting a sampling frequency according to the electricity utilization characteristics of a user, and collecting the power load information of the user according to the sampling frequency; uploading the power load information of the user to a database, analyzing the abnormal value of the uploaded power load information of the user, and performing early warning if the abnormality is found; and acquiring an abnormal value case, training a long-short-time memory network model according to the abnormal value case, and predicting whether the electric power is abnormal or not through the long-short-time memory network model. By the method and the corresponding system, dynamic supervision of the power load data is realized, and the response speed of the power load abnormal data is improved.
Description
Technical Field
The invention provides a power load remote monitoring and early warning method and system based on the Internet of things, and relates to the technical field of power load early warning.
Background
With the rapid development of economy and the improvement of the living standard of people, the power demand is continuously increased, and the stability and reliability of power supply become particularly important. However, the traditional method of the traditional power load monitoring and early warning method generally depends on periodic data reading, cannot realize real-time data update, so that the response to the sudden abnormal event is not timely enough, and even if the real-time data update is realized, the dynamic supervision of the data cannot be realized, so that the response to the sudden abnormal event of the power load is slow.
Disclosure of Invention
The invention provides a power load remote monitoring and early warning method and system based on the Internet of things, which are used for solving the problems mentioned above:
the invention provides a power load remote monitoring and early warning method based on the Internet of things, which comprises the following steps:
setting a sampling frequency according to the electricity utilization characteristics of a user, and collecting the power load information of the user according to the sampling frequency;
Uploading the power load information of the user to a database, analyzing the abnormal value of the uploaded power load information of the user, and performing early warning if the abnormality is found;
And acquiring an abnormal value case, training a long-short-time memory network model according to the abnormal value case, and predicting whether the electric power is abnormal or not through the long-short-time memory network model.
Further, setting a sampling frequency according to the electricity utilization characteristic of the user, and collecting the power load information of the user according to the sampling frequency, wherein the method comprises the following steps:
setting a sampling frequency according to a sampling frequency model according to the electricity utilization characteristics of a user, wherein the sampling frequency model is as follows: Wherein, Is the sampling frequency at which the sample is to be taken,Is the highest frequency in the acquired signal, alpha is a growth factor regulator, determines the maximum multiple increase of the variation response, 0 < alpha is less than or equal to 1,Is the change of the current observation value relative to the power use of the last observation period, beta is a curvature parameter used for controlling the growth speed of the function, beta is more than or equal to 0,Is the base power consumption value;
And installing current, voltage and power signal sensors at the power interface of the user, and acquiring power load information of the user according to the sampling frequency by the sensors.
Further, uploading the user power load information to a database, and performing outlier analysis on the uploaded user power load information, including:
uploading the user power load information to a database, wherein each information record is attached with a corresponding user identifier during uploading;
deleting and repairing damaged, lost and abnormal data points in the user power load information, and eliminating interference signals in the user power load information;
Acquiring a sliding window, calculating positive and negative abnormal values in the sliding window, acquiring an abnormal value range of the power load according to the positive and negative abnormal values, judging whether the current is abnormal according to the abnormal value range of the power load,
Specifically, the positive outlier determination formula is:
Wherein, Represents an abnormal accumulated value in the forward direction at the time point n,Represents the actual current value at the time point n, k represents the coefficient, the value range is (0, 1),Represents an abnormal accumulated value in the forward direction at the time point n-1, W represents a sliding window size, represents data considering the past W time points,A time current value at a time point n-i;
the negative outlier determination formula is:
Wherein, Represents the abnormal accumulated value of the negative direction at the time point n,Represents the actual current value at the time point n, k represents the coefficient, the value range is (0, 1),Represents an abnormal accumulated value in the negative direction at the time point n-1, W represents a sliding window size, and represents data considering the past W time points.
Further, acquiring the sliding window includes:
calculating a sliding window through a sliding window model, wherein the sliding window model is as follows:
where W is the sliding window size, Is a set standard rate of change, for normalizing the effect of the rate of change V,Is a small positive number set to avoid zero denominator,AndA lower limit and an upper limit of the window size, respectively, ensuring that the window size varies within this interval;
calculating the change rate of the current through a change rate model, wherein the change rate model is as follows:
wherein V represents the rate of change, AndRepresenting the current values at two consecutive time points, respectively.
Further, acquiring an outlier case, training a long-short-time memory network model according to the outlier case, and predicting whether the electric power is abnormal or not through the long-short-time memory network model, wherein the method comprises the following steps:
When the power load data is in an abnormal value range, early warning is carried out, meanwhile, power load abnormal data in a time window when the early warning is sent out are obtained, key features which are helpful for identifying abnormality in the power load abnormal data are extracted, and the key features comprise peak values, abrupt points and long-term trends of the power load;
Dividing the power load abnormal data into a training set and a testing set, training a long-short-time memory network model by using the data on the training set, and adjusting parameters to be optimal;
And receiving new power load data in real time, inputting the power load data into a pre-trained model, judging whether an abnormal mode exists according to the input real-time data by the long-short-time memory network model, detecting the abnormality, sending out early warning, and simultaneously taking the new early-warning power load abnormal data as training data to train the long-short-time memory network model.
The invention provides a power load remote monitoring and early warning system based on the Internet of things, which comprises the following components:
The power load information acquisition module is used for setting a sampling frequency according to the power utilization characteristics of a user and acquiring power load information of the user according to the sampling frequency;
The abnormality analysis module is used for uploading the power load information of the user to the database, analyzing the abnormal value of the uploaded power load information of the user, and carrying out early warning if the abnormality is found;
and the training anomaly prediction model module is used for acquiring an anomaly value case, training a long-short-time memory network model according to the anomaly value case, and predicting whether the electric power is anomaly or not through the long-short-time memory network model.
Further, the power load information acquisition module includes:
the sampling frequency setting module is used for setting the sampling frequency according to the sampling frequency model according to the electricity utilization characteristics of a user, and specifically, the sampling frequency model is as follows:
Wherein, Is the sampling frequency at which the sample is to be taken,Is the highest frequency in the acquired signal, alpha is a growth factor regulator, determines the maximum multiple increase of the variation response, 0 < alpha is less than or equal to 1,Is the change of the current observation value relative to the power use of the last observation period, beta is a curvature parameter used for controlling the growth speed of the function, beta is more than or equal to 0,Is the base power consumption value;
And the power load information acquisition module is used for installing current, voltage and power signal sensors at the power interface of the user according to the sampling frequency, and the sensors acquire the power load information of the user according to the sampling frequency.
Further, the anomaly analysis module includes:
The subsidiary subscriber identifier uploading database module is used for uploading the subscriber power load information to the database, and each information record is subsidiary with a corresponding subscriber identifier during uploading;
The data preprocessing module is used for deleting and repairing damaged, lost and abnormal data points in the user power load information and eliminating interference signals in the user power load information;
A judging data abnormality module for obtaining a sliding window, calculating positive and negative abnormal values in the sliding window, obtaining a power load abnormal value range according to the positive and negative abnormal values, judging whether the current is abnormal according to the power load abnormal value range,
Specifically, the positive outlier determination formula is:
Wherein, Represents the current abnormality accumulation value in the forward direction at the time point n,Represents the actual current value at the time point n, k represents the coefficient, the value range is (0, 1),Represents an abnormal accumulated value in the forward direction at the time point n-1, W represents a sliding window size, represents data considering the past W time points,A time current value at a time point n-i;
the negative outlier determination formula is:
Wherein, Represents the negative current anomaly cumulative value at time point n,Represents the actual current value at the time point n, k represents the coefficient, the value range is (0, 1),Represents an abnormal accumulated value in the negative direction at the time point n-1, W represents a sliding window size, and represents data considering the past W time points.
Further, the data anomaly judging module includes:
the sliding window acquisition module is used for calculating a sliding window through a sliding window model, and specifically, the sliding window model is as follows: where W is the sliding window size, Is a set standard rate of change, for normalizing the effect of the rate of change V,Is a small positive number set to avoid zero denominator,AndA lower limit and an upper limit of the window size, respectively, ensuring that the window size varies within this interval;
the change rate obtaining module is used for calculating the change rate of the current through a change rate model, and specifically, the change rate model is as follows:
wherein V represents the rate of change, AndRepresenting the current values at two consecutive time points, respectively.
Further, the training anomaly prediction model module includes:
The abnormal data acquisition module is used for carrying out early warning when the power load data is in an abnormal value range, simultaneously acquiring the power load abnormal data in a time window when the early warning is sent out, and extracting key features which are helpful for identifying the abnormality in the power load abnormal data, wherein the key features comprise peak values, abrupt points and long-term trends of the power load;
The training module is used for dividing the power load abnormal data into a training set and a testing set, training a long-time memory network model by using the data on the training set, and adjusting parameters to be optimal;
The real-time prediction abnormality module is used for receiving new power load data in real time, inputting the power load data into the pre-training model, judging whether an abnormality mode exists according to the input real-time data by the long-short-time memory network model, detecting abnormality, sending early warning, and simultaneously taking the new early-warning power load abnormality data as training data to train the long-short-time memory network model.
The invention has the beneficial effects that: the response speed is improved, the sampling frequency is dynamically adjusted, the response can be fast when the power use is changed significantly, key data can be recorded timely, and the potential problems can be recognized and responded more quickly; the accuracy and the reliability are improved, and the accuracy of predicting and detecting the power abnormality is improved by combining the abnormality detection with the LSTM model; the introduction of LSTM is particularly effective for those power modes that have complex periodicity and long-term dependence; the maintenance cost and risk are reduced, and the early warning system can help to find and solve problems in time and prevent larger property loss or safety accidents caused by the faults or abnormal operation of the electric equipment; the data-driven decision support, the collected detailed data and the deep analysis can provide data support for power system management, and help a decision maker to perform more accurate maintenance and optimization operation; by introducing a time window and positive and negative abnormal value models, dynamic supervision of power load data can be realized so as to respond to abnormal conditions of the power load data more quickly, and by the comprehensive technical scheme, intelligent and automatic power monitoring and management can be realized, and a high-efficiency and safe power use environment is provided.
Drawings
Fig. 1 is a schematic diagram of a remote monitoring and early warning method for electric load based on the internet of things.
Detailed Description
In order that the above-recited objects, features and advantages of the present application will be more clearly understood, a more particular description of the application will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings. It should be noted that, without conflict, the embodiments of the present application and features in the embodiments may be combined with each other.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, and the described embodiments are merely some, rather than all, embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.
The embodiment of the invention discloses a power load remote monitoring and early warning method based on the Internet of things, which comprises the following steps:
setting a sampling frequency according to the electricity utilization characteristics of a user, and collecting the power load information of the user according to the sampling frequency;
Uploading the power load information of the user to a database, analyzing the abnormal value of the uploaded power load information of the user, and performing early warning if the abnormality is found;
And acquiring an abnormal value case, training a long-short-time memory network model according to the abnormal value case, and predicting whether the electric power is abnormal or not through the long-short-time memory network model.
The working principle and the effect of the technical scheme are as follows: setting a sampling frequency, setting the sampling frequency of power data by using a model-based method according to the power utilization characteristics of a user, and dynamically adjusting the sampling rate by taking the highest frequency, the power utilization variation and the basic power consumption into consideration by the model, wherein the adjustment ensures that when the power utilization is obviously changed, the system can acquire data at a higher frequency, so that important power utilization condition change is captured; the data is collected and uploaded, and the power load information of the user is collected in real time and uploaded to a central database under the set sampling frequency, so that the data is not only up to date, but also can reflect the change of the power consumption mode in a finer granularity due to the adjustment of the sampling rate; outlier analysis, after data upload, to identify power usage that does not conform to conventional patterns, which may include sudden increases or decreases in power usage, which may be indicative of equipment failure, illegal access, or other potential problems; the early warning mechanism is started once an abnormal value is detected, and the early warning mechanism can inform a user or a maintenance team to check and deal with in time so as to avoid possible equipment damage or safety accidents; the long-short-time memory network model is used for training and predicting, the long-short-time memory network (LSTM) model is used for training according to historical outlier cases, the LSTM can process and predict time sequence data, long-term dependency relationship in the time sequence data is captured, the model learns what power use mode is likely to be followed by abnormality through training, in operation, the model can predict whether an abnormality mode occurs in the power data in real time, and possible problems are early warned. The response speed is improved, and key data can be recorded in time by dynamically adjusting the sampling frequency and rapidly responding when the power use is changed significantly. This helps to more quickly identify and respond to potential problems; and the accuracy and the reliability are improved, and the accuracy of predicting and detecting the power abnormality is improved by combining the abnormality detection with the LSTM model. The introduction of LSTM is particularly effective for those power modes that have complex periodicity and long-term dependence;
The maintenance cost and risk are reduced, and the early warning system can help to find and solve problems in time and prevent larger property loss or safety accidents caused by the faults or abnormal operation of the electric equipment; the data-driven decision support, the collected detailed data and the deep analysis can provide data support for power system management, and help a decision maker to perform more accurate maintenance and optimization operation; by the comprehensive technical scheme, more intelligent and automatic power monitoring and management can be realized, and a high-efficiency and safe power use environment is provided.
According to one embodiment of the invention, a remote monitoring and early warning method for electric load based on the Internet of things sets a sampling frequency according to the electricity utilization characteristics of a user, and acquires electric load information of the user according to the sampling frequency, and the method comprises the following steps:
setting a sampling frequency according to a sampling frequency model according to the electricity utilization characteristics of a user, wherein the sampling frequency model is as follows: Wherein, Is the sampling frequency at which the sample is to be taken,Is the highest frequency in the acquired signal, alpha is a growth factor regulator, determines the maximum multiple increase of the variation response, 0 < alpha is less than or equal to 1,Is the change of the current observation value relative to the power use of the last observation period, beta is a curvature parameter used for controlling the growth speed of the function, beta is more than or equal to 0,Is the base power consumption value;
And installing current, voltage and power signal sensors at the power interface of the user, and acquiring power load information of the user according to the sampling frequency by the sensors.
The working principle and the effect of the technical scheme are as follows: according to the technical scheme, a sensor is adopted to collect power load information (including current, voltage and power) of a user in real time, the data collection frequency is dynamically adjusted according to a preset digital model, the model is based on the change rate of current and previous power consumption, and the sampling frequency is adjusted by using a simplified logistic regression model, so that the power consumption is more sensitive to sudden changes of the power consumption, and the method aims at reducing data storage and processing requirements, and meanwhile, high responsiveness and monitoring capability on system state change are maintained; the adaptability: the sampling frequency can be dynamically adjusted according to the instant change of the power load, so that the system can rapidly respond and record events in detail when the power consumption is increased or reduced rapidly; data optimization reduces the data acquisition frequency in the period of stable power use, and optimizes the use of data storage and processing resources; the sensitivity and the instantaneity are realized, and the system can be optimized according to specific monitoring requirements by adjusting the parameters alpha and beta, so that the rapid response and the high sensitivity to mutation are ensured; cost-effective, dynamic adjustment of sampling frequency can reduce energy consumption and operation cost in low demand period, and simultaneously provide high quality data at critical moment, ensure economic operation and technical benefit of system, formulaDesigned to dynamically adjust the sampling frequency according to changes in power usage,This part ensures that the sampling frequency is at leastIs twice as large, satisfies the nyquist sampling theorem, prevents aliasing, and the entire model is simultaneously dependent on the relative changes in power usageAdjustments are made, alpha and beta allow the response sensitivity and growth rate to be tailored to different scenarios,Providing a smooth and nonlinear response power consumption change mode, which can excessively increase the adjustment frequency when the change is small and rapidly increase the frequency when the change is large, so as to ensure the capture of key data; the maximum value is taken between the calculated sampling frequency and 2000, so that the sampling frequency is not lower than 2000HZ, and the whole model can ensure that the lower the variation value is, the slower the sampling frequency is, the larger the variation value is, and the faster the sampling frequency is; the sampling frequency of the model can be flexibly adjusted according to the actual power change condition, so that resources are saved, and the model is more economical and efficient; the data loss is avoided, and the data loss or delay caused by insufficient sampling rate is reduced by ensuring that the sampling frequency is increased when the power load is changed greatly; noise and redundancy are reduced, sampling frequency is reduced when the power consumption is not changed greatly, data noise and redundancy can be reduced, and data quality and processing efficiency are improved; the adjustability, alpha and beta are introduced, so that the system can adjust the sensitivity and response speed according to different application requirements, and high customization capability is provided; in summary, the design method realizes technical flexibility and economic efficiency, effectively balances the relation between the monitoring requirement and the resource consumption, and is particularly suitable for large-scale power monitoring and management systems.
According to one embodiment of the invention, a remote monitoring and early warning method for power load based on the Internet of things is used for uploading user power load information to a database and analyzing an abnormal value of the uploaded user power load information, and comprises the following steps:
uploading the user power load information to a database, wherein each information record is attached with a corresponding user identifier during uploading;
deleting and repairing damaged, lost and abnormal data points in the user power load information, and eliminating interference signals in the user power load information;
Acquiring a sliding window, calculating positive and negative abnormal values in the sliding window, acquiring an abnormal value range of the power load according to the positive and negative abnormal values, judging whether the current is abnormal according to the abnormal value range of the power load,
Specifically, the positive outlier determination formula is:
Wherein, Represents an abnormal accumulated value in the forward direction at the time point n,Represents the actual current value at the time point n, k represents the coefficient, the value range is (0, 1),Represents an abnormal accumulated value in the forward direction at the time point n-1, W represents a sliding window size, represents data considering the past W time points,A time current value at a time point n-i;
the negative outlier determination formula is:
Wherein, Represents the abnormal accumulated value of the negative direction at the time point n,Represents the actual current value at the time point n, k represents the coefficient, the value range is (0, 1),Represents an abnormal accumulated value in the negative direction at the time point n-1, W represents a sliding window size, represents data considering the past W time points,，。
The working principle and the effect of the technical scheme are as follows: detecting current anomalies by monitoring and analyzing a system of power load information, receiving current data from a sensor by the system, uploading the data (with a user identifier) to a database, cleaning the data by using an advanced data processing technology, eliminating interference signals, calculating normal and abnormal ranges of current values by using a sliding window technology to identify potential abnormal current activities, and identifying the power anomalies by adopting a positive abnormal value and negative abnormal value accumulation method; the method not only detects the amplitude of the abnormality, but also accumulates the occurrence frequency and intensity of the abnormality, thereby providing a basis for more accurate judgment; sliding intra-window computationProviding a dynamic average of the current value over time, helping to track long-term trends; sliding intra-window computationThe fluctuation of the data is measured, and the larger the value is, the more severe the current value changes are; abnormal cumulative valueAndConfigured to accumulate outliers positively or negatively to help identify persistent abnormal activity, rather than occasional fluctuations; sensitivity and responsiveness can be enhanced, by accumulating outliers, the system can identify problems when the outliers begin to accumulate, rather than waiting until the outliers reach a high point; false alarms are reduced, normal ranges are defined based on historical data, and false alarms caused by normal fluctuation are reduced; the dynamic threshold value is used for dynamically calculating the k times value as the threshold value, so that the system can adapt to the natural fluctuation of data under different conditions, the adaptability of the system is stronger, and through the design, the system can improve the accuracy and efficiency of detecting the power load abnormality and provide guarantee for the safe and reliable operation of the power system; by defining positive and negative outlier accumulation valuesAndThe formula can take into account the persistence and cumulative effects of the anomaly event, meaning that even though a single data point may not be sufficient to trigger an anomaly alarm, a series of data points that continuously deviate from the normal range may cause the cumulative value to exceed a threshold value, triggering anomaly detection; Providing a starting point for the window size, the sliding window size will be the reference value, which is an adjustment factor, in the absence of other external factors such as the rate of change V, and the standard rate of change Working together, dynamically adjusting the size of the window based on real-time data, such design enabling the window size to be adaptively adjusted as the volatility of the power load increases or decreases; stability is maintained, and in this model,Also helps to ensure that the window size is not greatly adjusted due to temporary small fluctuation, thereby maintaining stability and continuity of analysis process, at presetWhen the value of (2) is selected according to the characteristics of the historical data, the analysis requirement and the change amplitude of the dataTypically, this is set by pre-experiments or empirically,Has a direct impact on the flexibility and responsiveness of the system, a smaller oneMaking the system more sensitive to changes, a largerIs more suitable for smoothing the long-term trend of a large amount of data,The key role of adjusting the size of the reference window is played in the sliding window model, so that the whole system can be flexibly adjusted according to the change condition of actual data, and the key role is particularly important to dynamic and complex data environments, such as application scenes of power load monitoring and the like. In summary, the improved formula design improves the accuracy, flexibility and robustness of the power load abnormality detection by introducing the concepts of the dynamic threshold and the average value of the sliding window, and the method can be better adapted to the dynamic characteristics of the data, effectively identify abnormal events and reduce the possibility of false alarm and missing report.
According to one embodiment of the invention, a power load remote monitoring and early warning method based on the Internet of things comprises the following steps:
calculating a sliding window through a sliding window model, wherein the sliding window model is as follows:
where W is the sliding window size, Is a set standard rate of change, for normalizing the effect of the rate of change V,Is a small positive number set to avoid zero denominator,AndA lower limit and an upper limit of the window size, respectively, ensuring that the window size varies within this interval;
calculating the change rate of the current through a change rate model, wherein the change rate model is as follows:
wherein V represents the rate of change, AndRepresenting the current values at two consecutive time points, respectively.
The working principle and the effect of the technical scheme are as follows: the sliding window model allows the sliding window size W to be dynamically adjusted according to the current change rate V, and when the change rate is large, the window is reduced to more accurately capture the current change; when the rate of change is smaller, the window is enlarged to contain more historical data, thereby improving stability and limiting the range, by setting the minimum window sizeAnd maximum window sizeCan ensure that the sliding window is not too small or too large, thereby avoiding erroneous judgment under extreme conditions, parametersThe existence of (2) is to prevent the denominator from becoming 0 when V is very close to 0, thereby ensuring mathematical stability of the formula; the sliding window model can adjust the window size according to the real-time change of the current data, so that the flexibility and the accuracy of anomaly detection are improved; by limiting the window size range, the interference of abnormal data on window size calculation is reduced, and the robustness of the algorithm is enhanced; the change rate model directly calculates the relative change rate of the current values of two adjacent time points, and can intuitively reflect the change condition of the current; the data of two adjacent time points are used for calculation, so that the complexity of data processing and calculation is simplified; the current change can be reflected rapidly, and the method is suitable for real-time monitoring and anomaly detection scenes; even if the current is slightly changed, the change rate model can be captured, so that the sensitivity of abnormality detection is improved; in summary, the design of these two models aims to improve the accuracy, sensitivity and robustness of the power load anomaly detection. By dynamically adjusting the size of the sliding window and calculating the current change rate in real time, the system can more effectively identify abnormal current fluctuation, so that corresponding measures are taken in time, and the stable operation of the power system is ensured.
According to one embodiment of the invention, an abnormal value case is obtained by a power load remote monitoring and early warning method based on the Internet of things, a long-short-time memory network model is trained according to the abnormal value case, and whether power is abnormal or not is predicted by the long-short-time memory network model, and the method comprises the following steps:
When the power load data is in an abnormal value range, early warning is carried out, meanwhile, power load abnormal data in a time window when the early warning is sent out are obtained, key features which are helpful for identifying abnormality in the power load abnormal data are extracted, and the key features comprise peak values, abrupt points and long-term trends of the power load;
Dividing the power load abnormal data into a training set and a testing set, training a long-short-time memory network model by using the data on the training set, and adjusting parameters to be optimal;
And receiving new power load data in real time, inputting the power load data into a pre-trained model, judging whether an abnormal mode exists according to the input real-time data by the long-short-time memory network model, detecting the abnormality, sending out early warning, and simultaneously taking the new early-warning power load abnormal data as training data to train the long-short-time memory network model.
The working principle and the effect of the technical scheme are as follows: the technical scheme is based on a long and short time memory network (LSTM), a deep learning model particularly suitable for processing time series data is used for monitoring power load data, and the LSTM can memorize long-term dependence information, so that the LSTM is particularly suitable for processing time series analysis and anomaly detection of the power data; data preparation and feature extraction, first extracting key features, such as peaks, discontinuities and long-term trends, from the electrical load data, which can help the model identify typical and abnormal patterns of the electrical load; dividing a data set, namely dividing the data into a training set and a testing set, wherein the training set is used for learning and parameter tuning of a model, and the testing set is used for verifying generalization capability and accuracy of the model; model training and optimization, training an LSTM model by using training set data, and achieving optimal performance by adjusting model parameters; during system operation, real-time power load data is input to the pre-trained LSTM model. If the model detects an abnormal mode, the system sends out early warning; the detected abnormal data are added into the training set, so that the model can perform self-optimization by continuously learning new data, and the accuracy and the efficiency of abnormality detection are improved. The recognition accuracy is improved, and the LSTM model can learn the complex mode and long-term dependence of the power load data, so that the abnormality can be accurately recognized in a changeable environment; the system can process the power data in real time and respond to abnormality quickly, thereby being beneficial to finding and solving the problems in the power system in time; self-adaptation and self-optimization, the model can adapt to new power load modes and environmental changes by continuously learning the latest abnormal data, and high-efficiency early warning performance is maintained; through continuous optimization and training, the LSTM model can more accurately understand what is the normal mode and the abnormal mode, and is helpful for reducing false alarms and false misses. Through the LSTM-based power load monitoring and abnormality detecting system, the operation state of the power facility can be efficiently and accurately monitored, and the reliability and the safety of the power system are improved.
According to one embodiment of the invention, a power load remote monitoring and early warning system based on the Internet of things comprises:
The power load information acquisition module is used for setting a sampling frequency according to the power utilization characteristics of a user and acquiring power load information of the user according to the sampling frequency;
The abnormality analysis module is used for uploading the power load information of the user to the database, analyzing the abnormal value of the uploaded power load information of the user, and carrying out early warning if the abnormality is found;
and the training anomaly prediction model module is used for acquiring an anomaly value case, training a long-short-time memory network model according to the anomaly value case, and predicting whether the electric power is anomaly or not through the long-short-time memory network model.
The working principle and the effect of the technical scheme are as follows: setting a sampling frequency, setting the sampling frequency of power data by using a model-based method according to the power utilization characteristics of a user, and dynamically adjusting the sampling rate by taking the highest frequency, the power utilization variation and the basic power consumption into consideration by the model, wherein the adjustment ensures that when the power utilization is obviously changed, the system can acquire data at a higher frequency, so that important power utilization condition change is captured; the data is collected and uploaded, and the power load information of the user is collected in real time and uploaded to a central database under the set sampling frequency, so that the data is not only up to date, but also can reflect the change of the power consumption mode in a finer granularity due to the adjustment of the sampling rate; outlier analysis, after data upload, to identify power usage that does not conform to conventional patterns, which may include sudden increases or decreases in power usage, which may be indicative of equipment failure, illegal access, or other potential problems; the early warning mechanism is started once an abnormal value is detected, and the early warning mechanism can inform a user or a maintenance team to check and deal with in time so as to avoid possible equipment damage or safety accidents; the long-short-time memory network model is used for training and predicting, the long-short-time memory network (LSTM) model is used for training according to historical outlier cases, the LSTM can process and predict time sequence data, long-term dependency relationship in the time sequence data is captured, the model learns what power use mode is likely to be followed by abnormality through training, in operation, the model can predict whether an abnormality mode occurs in the power data in real time, and possible problems are early warned. The response speed is improved, and key data can be recorded in time by dynamically adjusting the sampling frequency and rapidly responding when the power use is changed significantly. This helps to more quickly identify and respond to potential problems; and the accuracy and the reliability are improved, and the accuracy of predicting and detecting the power abnormality is improved by combining the abnormality detection with the LSTM model. The introduction of LSTM is particularly effective for those power modes that have complex periodicity and long-term dependence;
The maintenance cost and risk are reduced, and the early warning system can help to find and solve problems in time and prevent larger property loss or safety accidents caused by the faults or abnormal operation of the electric equipment; the data-driven decision support, the collected detailed data and the deep analysis can provide data support for power system management, and help a decision maker to perform more accurate maintenance and optimization operation; by the comprehensive technical scheme, more intelligent and automatic power monitoring and management can be realized, and a high-efficiency and safe power use environment is provided.
According to one embodiment of the invention, a power load remote monitoring and early warning system based on the Internet of things, wherein the power load information acquisition module comprises:
the sampling frequency setting module is used for setting the sampling frequency according to the sampling frequency model according to the electricity utilization characteristics of a user, and specifically, the sampling frequency model is as follows: Wherein, Is the sampling frequency at which the sample is to be taken,Is the highest frequency in the acquired signal, alpha is a growth factor regulator, determines the maximum multiple increase of the variation response, 0 < alpha is less than or equal to 1,Is the change of the current observation value relative to the power use of the last observation period, beta is a curvature parameter used for controlling the growth speed of the function, beta is more than or equal to 0,Is the base power consumption value;
And the power load information acquisition module is used for installing current, voltage and power signal sensors at the power interface of the user according to the sampling frequency, and the sensors acquire the power load information of the user according to the sampling frequency.
The working principle and the effect of the technical scheme are as follows: according to the technical scheme, a sensor is adopted to collect power load information (including current, voltage and power) of a user in real time, the data collection frequency is dynamically adjusted according to a preset digital model, the model is based on the change rate of current and previous power consumption, and the sampling frequency is adjusted by using a simplified logistic regression model, so that the power consumption is more sensitive to sudden changes of the power consumption, and the method aims at reducing data storage and processing requirements, and meanwhile, high responsiveness and monitoring capability on system state change are maintained; the adaptability: the sampling frequency can be dynamically adjusted according to the instant change of the power load, so that the system can rapidly respond and record events in detail when the power consumption is increased or reduced rapidly; data optimization reduces the data acquisition frequency in the period of stable power use, and optimizes the use of data storage and processing resources; the sensitivity and the instantaneity are realized, and the system can be optimized according to specific monitoring requirements by adjusting the parameters alpha and beta, so that the rapid response and the high sensitivity to mutation are ensured; cost-effective, dynamic adjustment of sampling frequency can reduce energy consumption and operation cost in low demand period, and simultaneously provide high quality data at critical moment, ensure economic operation and technical benefit of system, formulaDesigned to dynamically adjust the sampling frequency according to changes in power usage,This part ensures that the sampling frequency is at leastIs twice as large, satisfies the nyquist sampling theorem, prevents aliasing, and the entire model is simultaneously dependent on the relative changes in power usageAdjustments are made, alpha and beta allow the response sensitivity and growth rate to be tailored to different scenarios,Providing a smooth and nonlinear response power consumption change mode, which can excessively increase the adjustment frequency when the change is small and rapidly increase the frequency when the change is large, so as to ensure the capture of key data; the maximum value is taken between the calculated sampling frequency and 2000, so that the sampling frequency is not lower than 2000HZ, and the whole model can ensure that the lower the variation value is, the slower the sampling frequency is, the larger the variation value is, and the faster the sampling frequency is; the sampling frequency of the model can be flexibly adjusted according to the actual power change condition, so that resources are saved, and the model is more economical and efficient; the data loss is avoided, and the data loss or delay caused by insufficient sampling rate is reduced by ensuring that the sampling frequency is increased when the power load is changed greatly; noise and redundancy are reduced, sampling frequency is reduced when the power consumption is not changed greatly, data noise and redundancy can be reduced, and data quality and processing efficiency are improved; the adjustability, alpha and beta are introduced, so that the system can adjust the sensitivity and response speed according to different application requirements, and high customization capability is provided; in summary, the design method realizes technical flexibility and economic efficiency, effectively balances the relation between the monitoring requirement and the resource consumption, and is particularly suitable for large-scale power monitoring and management systems.
According to one embodiment of the invention, an electric load remote monitoring and early warning system based on the Internet of things, the abnormality analysis module comprises:
The subsidiary subscriber identifier uploading database module is used for uploading the subscriber power load information to the database, and each information record is subsidiary with a corresponding subscriber identifier during uploading;
The data preprocessing module is used for deleting and repairing damaged, lost and abnormal data points in the user power load information and eliminating interference signals in the user power load information;
A judging data abnormality module for obtaining a sliding window, calculating positive and negative abnormal values in the sliding window, obtaining a power load abnormal value range according to the positive and negative abnormal values, judging whether the current is abnormal according to the power load abnormal value range,
Specifically, the positive outlier determination formula is:
Wherein, Represents the current abnormality accumulation value in the forward direction at the time point n,Represents the actual current value at the time point n, k represents the coefficient, the value range is (0, 1),Represents an abnormal accumulated value in the forward direction at the time point n-1, W represents a sliding window size, represents data considering the past W time points,A time current value at a time point n-i;
the negative outlier determination formula is:
Wherein, Represents the negative current anomaly cumulative value at time point n,Represents the actual current value at the time point n, k represents the coefficient, the value range is (0, 1),Represents an abnormal accumulated value in the negative direction at the time point n-1, W represents a sliding window size, and represents data considering the past W time points.
The working principle and the effect of the technical scheme are as follows: detecting current anomalies by monitoring and analyzing a system of power load information, receiving current data from a sensor by the system, uploading the data (with a user identifier) to a database, cleaning the data by using an advanced data processing technology, eliminating interference signals, calculating normal and abnormal ranges of current values by using a sliding window technology to identify potential abnormal current activities, and identifying the power anomalies by adopting a positive abnormal value and negative abnormal value accumulation method; the method not only detects the amplitude of the abnormality, but also accumulates the occurrence frequency and intensity of the abnormality, thereby providing a basis for more accurate judgment; sliding intra-window computationProviding a dynamic average of the current value over time, helping to track long-term trends; sliding intra-window computationThe fluctuation of the data is measured, and the larger the value is, the more severe the current value changes are; abnormal cumulative valueAndConfigured to accumulate outliers positively or negatively to help identify persistent abnormal activity, rather than occasional fluctuations; sensitivity and responsiveness can be enhanced, by accumulating outliers, the system can identify problems when the outliers begin to accumulate, rather than waiting until the outliers reach a high point; false alarms are reduced, normal ranges are defined based on historical data, and false alarms caused by normal fluctuation are reduced; the dynamic threshold value is used for dynamically calculating the k times value as the threshold value, so that the system can adapt to the natural fluctuation of data under different conditions, the adaptability of the system is stronger, and through the design, the system can improve the accuracy and efficiency of detecting the power load abnormality and provide guarantee for the safe and reliable operation of the power system; by defining positive and negative outlier accumulation valuesAndThe formula can take into account the persistence and cumulative effects of the anomaly event, meaning that even though a single data point may not be sufficient to trigger an anomaly alarm, a series of data points that continuously deviate from the normal range may cause the cumulative value to exceed a threshold value, triggering anomaly detection; in summary, the improved formula design improves the accuracy, flexibility and robustness of the power load abnormality detection by introducing the concepts of the dynamic threshold and the average value of the sliding window, and the method can be better adapted to the dynamic characteristics of the data, effectively identify abnormal events and reduce the possibility of false alarm and missing report.
According to one embodiment of the invention, a power load remote monitoring and early warning system based on the Internet of things, the judging data abnormality module comprises:
the sliding window acquisition module is used for calculating a sliding window through a sliding window model, and specifically, the sliding window model is as follows: where W is the sliding window size, Is a set standard rate of change, for normalizing the effect of the rate of change V,Is a small positive number set to avoid zero denominator,AndA lower limit and an upper limit of the window size, respectively, ensuring that the window size varies within this interval;
the change rate obtaining module is used for calculating the change rate of the current through a change rate model, and specifically, the change rate model is as follows:
wherein V represents the rate of change, AndRepresenting the current values at two consecutive time points, respectively.
The working principle and the effect of the technical scheme are as follows: the sliding window model allows the sliding window size W to be dynamically adjusted according to the current change rate V, and when the change rate is large, the window is reduced to more accurately capture the current change; when the rate of change is smaller, the window is enlarged to contain more historical data, thereby improving stability and limiting the range, by setting the minimum window sizeAnd maximum window sizeCan ensure that the sliding window is not too small or too large, thereby avoiding erroneous judgment under extreme conditions, parametersThe existence of (2) is to prevent the denominator from becoming 0 when V is very close to 0, thereby ensuring mathematical stability of the formula; the sliding window model can adjust the window size according to the real-time change of the current data, so that the flexibility and the accuracy of anomaly detection are improved; by limiting the window size range, the interference of abnormal data on window size calculation is reduced, and the robustness of the algorithm is enhanced; the change rate model directly calculates the relative change rate of the current values of two adjacent time points, and can intuitively reflect the change condition of the current; the data of two adjacent time points are used for calculation, so that the complexity of data processing and calculation is simplified; the current change can be reflected rapidly, and the method is suitable for real-time monitoring and anomaly detection scenes; even if the current is slightly changed, the change rate model can be captured, so that the sensitivity of abnormality detection is improved; in summary, the design of these two models aims to improve the accuracy, sensitivity and robustness of the power load anomaly detection. By dynamically adjusting the size of the sliding window and calculating the current change rate in real time, the system can more effectively identify abnormal current fluctuation, so that corresponding measures are taken in time, and the stable operation of the power system is ensured.
According to one embodiment of the invention, a power load remote monitoring and early warning system based on the Internet of things, the training abnormality prediction model module comprises:
The abnormal data acquisition module is used for carrying out early warning when the power load data is in an abnormal value range, simultaneously acquiring the power load abnormal data in a time window when the early warning is sent out, and extracting key features which are helpful for identifying the abnormality in the power load abnormal data, wherein the key features comprise peak values, abrupt points and long-term trends of the power load;
The training module is used for dividing the power load abnormal data into a training set and a testing set, training a long-time memory network model by using the data on the training set, and adjusting parameters to be optimal;
The real-time prediction abnormality module is used for receiving new power load data in real time, inputting the power load data into the pre-training model, judging whether an abnormality mode exists according to the input real-time data by the long-short-time memory network model, detecting abnormality, sending early warning, and simultaneously taking the new early-warning power load abnormality data as training data to train the long-short-time memory network model.
The working principle and the effect of the technical scheme are as follows: setting a sampling frequency, setting the sampling frequency of power data by using a model-based method according to the power utilization characteristics of a user, and dynamically adjusting the sampling rate by taking the highest frequency, the power utilization variation and the basic power consumption into consideration by the model, wherein the adjustment ensures that when the power utilization is obviously changed, the system can acquire data at a higher frequency, so that important power utilization condition change is captured; the data is collected and uploaded, and the power load information of the user is collected in real time and uploaded to a central database under the set sampling frequency, so that the data is not only up to date, but also can reflect the change of the power consumption mode in a finer granularity due to the adjustment of the sampling rate; outlier analysis, after data upload, to identify power usage that does not conform to conventional patterns, which may include sudden increases or decreases in power usage, which may be indicative of equipment failure, illegal access, or other potential problems; the early warning mechanism is started once an abnormal value is detected, and the early warning mechanism can inform a user or a maintenance team to check and deal with in time so as to avoid possible equipment damage or safety accidents; the long-short-time memory network model is used for training and predicting, the long-short-time memory network (LSTM) model is used for training according to historical outlier cases, the LSTM can process and predict time sequence data, long-term dependency relationship in the time sequence data is captured, the model learns what power use mode is likely to be followed by abnormality through training, in operation, the model can predict whether an abnormality mode occurs in the power data in real time, and possible problems are early warned. The response speed is improved, and key data can be recorded in time by dynamically adjusting the sampling frequency and rapidly responding when the power use is changed significantly. This helps to more quickly identify and respond to potential problems; and the accuracy and the reliability are improved, and the accuracy of predicting and detecting the power abnormality is improved by combining the abnormality detection with the LSTM model. The introduction of LSTM is particularly effective for those power modes that have complex periodicity and long-term dependence;
The maintenance cost and risk are reduced, and the early warning system can help to find and solve problems in time and prevent larger property loss or safety accidents caused by the faults or abnormal operation of the electric equipment; the data-driven decision support, the collected detailed data and the deep analysis can provide data support for power system management, and help a decision maker to perform more accurate maintenance and optimization operation; by the comprehensive technical scheme, more intelligent and automatic power monitoring and management can be realized, and a high-efficiency and safe power use environment is provided.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (10)
1. The utility model provides a power load remote monitoring early warning method based on thing networking which characterized in that, the method includes:
setting a sampling frequency according to the electricity utilization characteristics of a user, and collecting the power load information of the user according to the sampling frequency;
Uploading the power load information of the user to a database, analyzing the abnormal value of the uploaded power load information of the user, and performing early warning if the abnormality is found;
And acquiring an abnormal value case, training a long-short-time memory network model according to the abnormal value case, and predicting whether the electric power is abnormal or not through the long-short-time memory network model.
2. The method for remotely monitoring and early warning the electric load based on the Internet of things according to claim 1, wherein the method is characterized in that a sampling frequency is set according to the electricity utilization characteristics of a user, and the electric load information of the user is collected according to the sampling frequency, and the method comprises the following steps:
setting a sampling frequency according to a sampling frequency model according to the electricity utilization characteristics of a user, wherein the sampling frequency model is as follows: Wherein, Is the sampling frequency at which the sample is to be taken,Is the highest frequency in the acquired signal, alpha is a growth factor regulator, determines the maximum multiple increase of the variation response, 0 < alpha is less than or equal to 1,Is the change of the current observation value relative to the power use of the last observation period, beta is a curvature parameter used for controlling the growth speed of the function, beta is more than or equal to 0,Is the base power consumption value;
And installing current, voltage and power signal sensors at the power interface of the user, and acquiring power load information of the user according to the sampling frequency by the sensors.
3. The method for remote monitoring and early warning of electric power load based on the internet of things according to claim 1, wherein uploading the information of the electric power load of the user to a database, and analyzing the abnormal value of the uploaded information of the electric power load of the user comprises the following steps:
uploading the user power load information to a database, wherein each information record is attached with a corresponding user identifier during uploading;
deleting and repairing damaged, lost and abnormal data points in the user power load information, and eliminating interference signals in the user power load information;
Acquiring a sliding window, calculating positive and negative abnormal values in the sliding window, acquiring an abnormal value range of the power load according to the positive and negative abnormal values, judging whether the current is abnormal according to the abnormal value range of the power load,
Specifically, the positive outlier determination formula is:
Wherein, Represents an abnormal accumulated value in the forward direction at the time point n,Represents the actual current value at the time point n, k represents the coefficient, the value range is (0, 1),Represents an abnormal accumulated value in the forward direction at the time point n-1, W represents a sliding window size, represents data considering the past W time points,A time current value at a time point n-i;
the negative outlier determination formula is:
Wherein, Represents the abnormal accumulated value of the negative direction at the time point n,Represents the actual current value at the time point n, k represents the coefficient, the value range is (0, 1),Represents an abnormal accumulated value in the negative direction at the time point n-1, W represents a sliding window size, and represents data considering the past W time points.
4. The method for remotely monitoring and early warning power load based on the internet of things according to claim 3, wherein the step of acquiring the sliding window comprises the following steps:
calculating a sliding window through a sliding window model, wherein the sliding window model is as follows:
where W is the sliding window size, The reference size, representing the sliding window, is a preset value,Is a set standard rate of change, for normalizing the effect of the rate of change V,Is a small positive number set to avoid zero denominator,AndA lower limit and an upper limit of the window size, respectively, ensuring that the window size varies within this interval;
calculating the change rate of the current through a change rate model, wherein the change rate model is as follows:
wherein V represents the rate of change, AndRepresenting the current values at two consecutive time points, respectively.
5. The method for remotely monitoring and early warning of electric power load based on the internet of things according to claim 1, wherein obtaining an abnormal value case, training a long-short-time memory network model according to the abnormal value case, and predicting whether electric power is abnormal or not through the long-short-time memory network model comprises:
When the power load data is in an abnormal value range, early warning is carried out, meanwhile, power load abnormal data in a time window when the early warning is sent out are obtained, key features which are helpful for identifying abnormality in the power load abnormal data are extracted, and the key features comprise peak values, abrupt points and long-term trends of the power load;
Dividing the power load abnormal data into a training set and a testing set, training a long-short-time memory network model by using the data on the training set, and adjusting parameters to be optimal;
And receiving new power load data in real time, inputting the power load data into a pre-trained model, judging whether an abnormal mode exists according to the input real-time data by the long-short-time memory network model, detecting the abnormality, sending out early warning, and simultaneously taking the new early-warning power load abnormal data as training data to train the long-short-time memory network model.
6. Electric power load remote monitoring early warning system based on thing networking, its characterized in that, the system includes:
The power load information acquisition module is used for setting a sampling frequency according to the power utilization characteristics of a user and acquiring power load information of the user according to the sampling frequency;
The abnormality analysis module is used for uploading the power load information of the user to the database, analyzing the abnormal value of the uploaded power load information of the user, and carrying out early warning if the abnormality is found;
and the training anomaly prediction model module is used for acquiring an anomaly value case, training a long-short-time memory network model according to the anomaly value case, and predicting whether the electric power is anomaly or not through the long-short-time memory network model.
7. The power load remote monitoring and early warning system based on the internet of things according to claim 6, wherein the power load information acquisition module comprises:
the sampling frequency setting module is used for setting the sampling frequency according to the sampling frequency model according to the electricity utilization characteristics of a user, and specifically, the sampling frequency model is as follows: Wherein, Is the sampling frequency at which the sample is to be taken,Is the highest frequency in the acquired signal, alpha is a growth factor regulator, determines the maximum multiple increase of the variation response, 0 < alpha is less than or equal to 1,Is the change of the current observation value relative to the power use of the last observation period, beta is a curvature parameter used for controlling the growth speed of the function, beta is more than or equal to 0,Is the base power consumption value;
And the power load information acquisition module is used for installing current, voltage and power signal sensors at the power interface of the user according to the sampling frequency, and the sensors acquire the power load information of the user according to the sampling frequency.
8. The power load remote monitoring and early warning system based on the internet of things of claim 6, wherein the anomaly analysis module comprises:
The subsidiary subscriber identifier uploading database module is used for uploading the subscriber power load information to the database, and each information record is subsidiary with a corresponding subscriber identifier during uploading;
The data preprocessing module is used for deleting and repairing damaged, lost and abnormal data points in the user power load information and eliminating interference signals in the user power load information;
A judging data abnormality module for obtaining a sliding window, calculating positive and negative abnormal values in the sliding window, obtaining a power load abnormal value range according to the positive and negative abnormal values, judging whether the current is abnormal according to the power load abnormal value range,
Specifically, the positive outlier determination formula is:
Wherein, Represents an abnormal accumulated value in the forward direction at the time point n,Represents the actual current value at the time point n, k represents the coefficient, the value range is (0, 1),Represents an abnormal accumulated value in the forward direction at the time point n-1, W represents a sliding window size, represents data considering the past W time points,A time current value at a time point n-i;
the negative outlier determination formula is:
Wherein, Represents the negative current anomaly cumulative value at time point n,Represents the actual current value at the time point n, k represents the coefficient, the value range is (0, 1),Represents an abnormal accumulated value in the negative direction at the time point n-1, W represents a sliding window size, and represents data considering the past W time points.
9. The power load remote monitoring and early warning system based on the internet of things according to claim 8, wherein the judging data abnormality module comprises:
the sliding window acquisition module is used for calculating a sliding window through a sliding window model, and specifically, the sliding window model is as follows: where W is the sliding window size, Is a set standard rate of change, for normalizing the effect of the rate of change V,Is a small positive number set to avoid zero denominator,AndA lower limit and an upper limit of the window size, respectively, ensuring that the window size varies within this interval;
the change rate obtaining module is used for calculating the change rate of the current through a change rate model, and specifically, the change rate model is as follows:
wherein V represents the rate of change, AndRepresenting the current values at two consecutive time points, respectively.
10. The power load remote monitoring and early warning system based on the internet of things of claim 6, wherein the training anomaly prediction model module comprises:
The abnormal data acquisition module is used for carrying out early warning when the power load data is in an abnormal value range, simultaneously acquiring the power load abnormal data in a time window when the early warning is sent out, and extracting key features which are helpful for identifying the abnormality in the power load abnormal data, wherein the key features comprise peak values, abrupt points and long-term trends of the power load;
The training module is used for dividing the power load abnormal data into a training set and a testing set, training a long-time memory network model by using the data on the training set, and adjusting parameters to be optimal;
The real-time prediction abnormality module is used for receiving new power load data in real time, inputting the power load data into the pre-training model, judging whether an abnormality mode exists according to the input real-time data by the long-short-time memory network model, detecting abnormality, sending early warning, and simultaneously taking the new early-warning power load abnormality data as training data to train the long-short-time memory network model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410980734.4A CN118539616B (en) | 2024-07-22 | 2024-07-22 | Power load remote monitoring and early warning method and system based on Internet of things |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410980734.4A CN118539616B (en) | 2024-07-22 | 2024-07-22 | Power load remote monitoring and early warning method and system based on Internet of things |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118539616A true CN118539616A (en) | 2024-08-23 |
| CN118539616B CN118539616B (en) | 2024-09-27 |
Family
ID=92389903
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410980734.4A Active CN118539616B (en) | 2024-07-22 | 2024-07-22 | Power load remote monitoring and early warning method and system based on Internet of things |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118539616B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2020035413A (en) * | 2018-08-28 | 2020-03-05 | 日鉄エンジニアリング株式会社 | Electric power demand prediction system, construction method of electric power demand prediction model, program, business support system |
| CN113572149A (en) * | 2021-04-28 | 2021-10-29 | 湖北工业大学 | Overload device early warning system based on non-invasive power load monitoring |
| CN117458713A (en) * | 2023-10-27 | 2024-01-26 | 贵州电网有限责任公司 | Monitoring management method and system based on power distribution automation |
| CN117595231A (en) * | 2023-10-20 | 2024-02-23 | 国网安徽省电力有限公司六安供电公司 | Intelligent power grid distribution management system and method thereof |
| CN117977536A (en) * | 2023-10-26 | 2024-05-03 | 北京中电普华信息技术有限公司 | Smart power grid load identification method |
-
2024
- 2024-07-22 CN CN202410980734.4A patent/CN118539616B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2020035413A (en) * | 2018-08-28 | 2020-03-05 | 日鉄エンジニアリング株式会社 | Electric power demand prediction system, construction method of electric power demand prediction model, program, business support system |
| CN113572149A (en) * | 2021-04-28 | 2021-10-29 | 湖北工业大学 | Overload device early warning system based on non-invasive power load monitoring |
| CN117595231A (en) * | 2023-10-20 | 2024-02-23 | 国网安徽省电力有限公司六安供电公司 | Intelligent power grid distribution management system and method thereof |
| CN117977536A (en) * | 2023-10-26 | 2024-05-03 | 北京中电普华信息技术有限公司 | Smart power grid load identification method |
| CN117458713A (en) * | 2023-10-27 | 2024-01-26 | 贵州电网有限责任公司 | Monitoring management method and system based on power distribution automation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118539616B (en) | 2024-09-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113305645B (en) | Numerical control machine tool cutter residual life prediction method based on hybrid neural model | |
| CN117113729A (en) | Digital twinning-based power equipment online state monitoring system | |
| CN109492790A (en) | Wind turbines health control method based on neural network and data mining | |
| CN117273402A (en) | Energy-saving management system and method for glass deep processing production line based on Internet of things technology | |
| CN117521800A (en) | Intelligent power transmission line inspection method based on knowledge graph | |
| CN117331790A (en) | Machine room fault detection method and device for data center | |
| CN118539616B (en) | Power load remote monitoring and early warning method and system based on Internet of things | |
| CN118408583B (en) | Encoder fault diagnosis method and system | |
| CN118157327B (en) | Cable sheath circulation monitoring device based on NB-IOT technology | |
| CN118155047B (en) | Small battery management method suitable for alpine region | |
| CN118169586A (en) | Self-wake-up diagnosis method and system for battery management | |
| CN117688464A (en) | Hidden danger analysis method and system based on multi-source sensor data | |
| CN118112937A (en) | Intelligent control system and method for micro-positive pressure of farm based on multisource fusion perception | |
| CN117743961A (en) | Cable joint fault detection method and system based on long-short-term memory LSTM | |
| CN117494009A (en) | Electrical equipment state evaluation method based on insulating material pyrolysis analysis and cloud platform | |
| CN118520234B (en) | I-type concentrator terminal with real-time data monitoring function | |
| CN118534829B (en) | Method and system for monitoring full life cycle of liquid flow energy storage equipment based on Internet of things | |
| CN118332476B (en) | Fault diagnosis method, system, equipment and storage medium for gateway metering device of factory station | |
| CN117034157B (en) | Hydropower equipment fault identification method and system combining multimodal operation data | |
| CN118405025B (en) | Fire engine battery management system | |
| CN118643681B (en) | Overload monitoring method and system for diesel generator | |
| CN117990976B (en) | Intelligent ammeter electricity consumption monitoring method and system | |
| CN211086964U (en) | Remote equipment vibration management system based on Internet of things | |
| CN118054753A (en) | Photovoltaic information monitoring method, system, equipment and medium | |
| CN118411846A (en) | Intelligent algorithm-based traffic signal lamp fault automatic detection and diagnosis method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |