CN118311362B

CN118311362B - Running state monitoring method for energy-saving medium-power direct-current speed regulating device

Info

Publication number: CN118311362B
Application number: CN202410733863.3A
Authority: CN
Inventors: 吴云龙; 刘宗钦; 陈家豪; 沈建华
Original assignee: Huzhou Jiwei Electronics Technology Co ltd
Current assignee: Huzhou Jiwei Electronics Technology Co ltd
Priority date: 2024-06-07
Filing date: 2024-06-07
Publication date: 2024-08-09
Anticipated expiration: 2044-06-07
Also published as: CN118311362A

Abstract

The invention relates to the technical field of data anomaly monitoring, in particular to an operation state monitoring method for an energy-saving medium-power direct current speed regulating device. According to the invention, through acquiring various performance time sequence data of the energy-saving medium-power direct-current speed regulating device, the data change characteristic value at each data point in each performance time sequence data is determined based on the numerical characteristic and the change trend of each time under each performance time sequence data. Meanwhile, the correlation characteristic values among the performance time sequence data are determined by considering the difference of the data change characteristic values among different performance time sequence data. And then determining the nearest neighbor distance of each moment based on the correlation characteristic values among the time sequence data with different performances and the data change characteristic values of the data points under the time sequence data with different performances, so that the abnormal moment is detected more accurately. The method combines the relevance between the local relation and the multidimensional data, and effectively improves the accuracy of monitoring the abnormality of the COF algorithm under the multidimensional data.

Description

Running state monitoring method for energy-saving medium-power direct-current speed regulating device

Technical Field

The invention relates to the technical field of data anomaly monitoring, in particular to an operation state monitoring method for an energy-saving medium-power direct current speed regulating device.

Background

The energy-saving medium-power direct current speed regulating device combines the characteristics of energy-saving technology and medium-power direct current speed regulation, and the control circuit is responsible for generating regulating signals to regulate the voltage and the current output by the power circuit, so that the accurate control of the rotating speed of the direct current motor is realized; and the output is dynamically adjusted according to the actual needs, so that unnecessary energy consumption is reduced, and the operation cost is reduced. Therefore, the running state of the direct current speed regulating device needs to be monitored when the direct current speed regulating device runs, and the running condition of the direct current speed regulating device is regulated according to the obtained running state monitoring result, so that the working efficiency is improved, and the waste of energy sources is avoided.

In the prior art, when abnormal monitoring is performed on the running state of the energy-saving medium-power direct current speed regulator, a COF algorithm is generally adopted to perform performance data, such as: the voltage, the current, the load, the motor rotation speed and the like are subjected to anomaly analysis so as to determine the anomaly time, however, the COF algorithm usually adopts the euclidean distance as a judgment condition when determining the nearest neighbor distance of data, and different state data have the mutual influence, for example, the current and the load quality inspection have certain correlation, the current can correspondingly change when the load changes, and meanwhile, the energy-saving medium-power direct-current speed regulating device usually needs to comprehensively analyze various performance data when dynamically regulating, so that the condition in COF anomaly monitoring is carried out by using the euclidean distance in only one performance data, and the accuracy of the final anomaly monitoring result is lower.

Disclosure of Invention

In order to solve the technical problem that the final abnormal monitoring result accuracy of the energy-saving medium-power direct current speed regulator is lower when only using Euclidean distance as the condition in COF abnormal monitoring in one performance data, the invention aims to provide an operation state monitoring method for the energy-saving medium-power direct current speed regulator, which adopts the following technical scheme:

Acquiring various performance time sequence data of the energy-saving medium-power direct current speed regulating device in the same time period when the energy-saving medium-power direct current speed regulating device operates, and acquiring the data change characteristics of each data point in each performance time sequence data;

In a preset neighborhood corresponding to each data point, determining an association weight corresponding to each data point in each performance time sequence data based on the difference condition of the data change characteristics of the data points under different performance time sequence data; under any two different performance time sequence data, analyzing the difference of the data change characteristics of all the data points, and combining the association weights of the data points to obtain association characteristics between any two different performance time sequence data;

Determining the nearest neighbor distance corresponding to each moment based on the association characteristics among the time sequence data of different performances and the data change characteristics of the data points of different moments under the time sequence data of different performances; based on the nearest neighbor distance at each moment, the COF algorithm is combined to monitor the abnormal operation of the energy-saving medium-power direct current speed regulating device.

Further, the plurality of performance timing data includes at least: current timing data, voltage timing data, motor speed timing data, and load timing data.

Further, the acquiring the data change characteristic of each data point in each performance time series data includes:

Fusing the numerical characteristics of the data points corresponding to each moment under each performance time sequence data, and determining the performance characteristic value corresponding to each moment; and selecting one kind of performance time sequence data as the data to be tested, and determining the data change characteristic value of each data point in the data to be tested based on the fluctuation condition of the numerical values of the data points in the data to be tested and combining the difference condition of the performance characteristic values between the corresponding moments of the data points.

Further, the method for acquiring the data change characteristic value comprises the following steps:

In the data to be measured, for any one data point;

determining a trend feature value corresponding to the data point based on a difference between the performance feature value at the time corresponding to the data point and the performance feature value at the time corresponding to the data point adjacent in time sequence;

obtaining a first fluctuation parameter at the data point based on the data change degree of the data point and the adjacent data point;

Taking the variance of the first fluctuation parameters of all the data points as the second fluctuation parameters of the data points in a preset window corresponding to the data points;

and obtaining a data change characteristic value of the data point under the data to be measured according to the trend characteristic value, the first fluctuation parameter and the second fluctuation parameter of the data point, wherein the trend characteristic value, the first fluctuation parameter and the second fluctuation parameter are positively correlated with the data change characteristic value.

Further, the method for acquiring the association weight comprises the following steps:

in the data to be measured, one data point is selected as a point to be measured, the moment corresponding to the point to be measured is taken as the moment to be measured, and the moment of all the data points in the preset neighborhood corresponding to the point to be measured is taken as the reference moment;

taking other performance time sequence data except the data to be measured as comparison data; for any one comparison data, determining the maximum association factor of the to-be-measured point under the comparison data based on the difference between the to-be-measured point and the data change characteristic values of the data points corresponding to all the reference moments in the comparison data;

and carrying out averaging treatment on the maximum association factors of the points to be measured under all the comparison data to obtain the association weights of the points to be measured in the data to be measured.

Further, the analyzing the difference of the data change characteristics of all the data points under any two different performance time sequence data and combining the association weights of the data points to obtain the association characteristics between any two different performance time sequence data includes:

Combining any two performance time sequence data to obtain all non-repeated data combinations;

In each data combination, one of the performance time sequence data is used as first data, and the other performance time sequence data is used as second data;

optionally taking a moment as a target moment, fusing the difference condition of the data change characteristic value of the corresponding data point in the first data at the target moment and the data change characteristic value of the corresponding data point in the second data at the target moment with the association weight of the corresponding data point in the first data and the second data at the target moment, and determining the association characteristic factors of the first data and the second data at the target moment;

and carrying out averaging treatment on the correlation characteristic value factors of the first data and the second data at all moments to obtain the correlation characteristic value between the first data and the second data.

Further, the method for acquiring the association characteristic factors comprises the following steps:

Taking the difference between the data change characteristic value of the corresponding data point in the first data at the target moment and the data change characteristic value of the corresponding data point in the second data at the target moment as a difference factor, and taking the value obtained after carrying out negative correlation mapping and normalization on the difference factor as a first adjustment factor of the first data and the second data at the target moment;

Taking the average value of the association weight of the corresponding data point in the first data at the target moment and the association weight of the corresponding data point in the second data at the target moment as a second adjustment factor;

And taking the product of the first adjustment factor and the second adjustment factor as an associated characteristic factor of the first data and the second data at the target time.

Further, the nearest neighbor distance obtaining method includes:

Taking other moments except the target moment as comparison moments, and taking the next preset number of moments adjacent in time sequence of the target moment as moments to be analyzed;

Carrying out change consistency analysis on the difference of the data change characteristic values, the numerical characteristic and the association characteristic value between two performance time sequence data in the data combination at the target moment and all the comparison moments in different data combinations to obtain a distance factor at the target moment;

carrying out change consistency analysis on the difference of the characteristic values of the data change in different data combinations at the moment to be analyzed and all subsequent moments in time sequence, the numerical characteristic and the associated characteristic value between two performance time sequence data in the data combination to obtain a distance factor of the moment to be analyzed;

and taking the sum value of the distance factor of the target moment and the distance factors of all corresponding moments to be analyzed as the nearest neighbor distance of the target moment.

Further, the process of obtaining the distance factor through the change consistency analysis comprises the following steps:

Determining a first change parameter of the target time and each comparison time under the first data in each data combination based on the difference of the data change characteristic values of the target time and each comparison time and the numerical value difference; determining a second change parameter of the target time and each comparison time under the second data based on the difference of the data change characteristic values of the target time and each comparison time in the second data and the numerical value difference; determining a change consistency factor of the target moment and each comparison moment based on the first change parameter, the second change parameter and the correlation characteristic value between the first data and the second data; normalizing the sum of the change consistency factors of the target moment and each comparison moment under all data combinations to obtain a distance parameter; taking the maximum value of the distance parameters of the target time and all the comparison time as a distance factor of the target time;

For any moment to be analyzed, taking all moments after the moment to be analyzed in time sequence as reference moments; determining a first change parameter of the time to be analyzed and each reference time under the first data based on the difference of the characteristic value of the data change of the time to be analyzed and each reference time in the first data and the numerical value difference; determining a second change parameter of the time to be analyzed and each reference time under the second data based on the difference of the characteristic value of the data change of the time to be analyzed and each reference time in the second data and the numerical value difference; determining a change consistency factor of the moment to be analyzed and each reference moment based on the first change parameter, the second change parameter and the correlation characteristic value between the first data and the second data; normalizing the sum of the change consistency factors of the moment to be analyzed and each reference moment under all data combinations to obtain a distance parameter; and taking the maximum value of the distance parameters of the moment to be analyzed and all the reference moments as a distance factor of the moment to be analyzed.

Further, the method for obtaining the variation consistency factor comprises the following steps:

taking the difference between the first variation parameter and the second variation parameter as a variation parameter difference value;

And carrying out negative correlation mapping on the variation parameter difference value and taking the product of the normalized value and the correlation characteristic value as the variation consistency factor.

The invention has the following beneficial effects:

Because of a certain correlation among various performance data, the invention firstly acquires various performance time sequence data and the data change characteristics of each data point in each time sequence data in the operation process of the energy-saving type medium-power direct-current speed regulating device under the same time period, and the data change characteristics can better reflect the dynamic change of the operation state of the energy-saving type medium-power direct-current speed regulating device and provide a data basis for subsequent abnormal monitoring. Further, the change condition of the data in the local range is analyzed, so that a preset neighborhood corresponding to each data point is determined, the difference condition of the change characteristics of the data among the data points under different performance time sequence data is analyzed in the neighborhood, the association weight of the data points is determined, then the difference of the change characteristics of the data points in the different performance time sequence data is combined, the association characteristics among the different performance time sequence data are analyzed, and the association characteristics characterize the association among the different performance time sequence data. Then, based on the association characteristics among different performance time sequence data and the data change characteristics of the data points under different performance time sequence data, the nearest neighbor distance corresponding to each moment is determined, and the method for determining the nearest neighbor distance not only considers the local relation among the data points, but also combines the relativity among different performance time sequence data, so that the moment when the power direct current speed regulating device in the energy-saving type is abnormal can be detected more accurately in the subsequent process, the operation abnormality monitoring of the power direct current speed regulating device in the energy-saving type is carried out on the basis of a COF algorithm, the moment when the abnormality occurs can be screened more accurately, and the accuracy of the abnormality monitoring is improved.

Drawings

In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.

FIG. 1 is a flow chart of a method for monitoring the operation status of an energy-saving medium-power DC speed regulating device according to an embodiment of the present invention;

FIG. 2 is a flow chart of an analysis process of a data change feature according to an embodiment of the present invention;

FIG. 3 is a flowchart of a method for obtaining a characteristic value of a data change according to an embodiment of the present invention;

FIG. 4 is a schematic diagram of a system for monitoring an operation status of an energy-saving medium-power DC speed regulating device according to an embodiment of the present invention;

FIG. 5 is a system block diagram of an operational status monitoring system for an energy efficient medium power DC speed governor according to one embodiment of the present invention;

fig. 6 is a schematic diagram of a computer-readable storage medium according to an embodiment of the present invention.

Detailed Description

In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following is a specific implementation, structure, characteristics and effects of an operation state monitoring method for an energy-saving medium-power direct current speed regulating device according to the invention, which are described in detail below with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

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 following specifically describes a specific scheme of the running state monitoring method for the energy-saving medium-power direct current speed regulating device provided by the invention with reference to the accompanying drawings.

Referring to fig. 1, a flowchart of a method for monitoring an operation state of an energy-saving medium-power dc speed regulating device according to an embodiment of the invention is shown, and the method includes the following steps:

Step S1: and acquiring various performance time sequence data of the energy-saving medium-power direct current speed regulating device in the same time period, and acquiring the data change characteristics of each data point in each performance time sequence data.

The energy-saving medium-power direct current speed regulating device combines the characteristics of energy-saving technology and medium-power direct current speed regulation, and aims to improve the energy use efficiency and meet the requirements of medium-power application. The operation state of the device is required to be monitored during operation, and the operation condition of the device is adjusted according to the obtained operation state monitoring result so as to improve the working efficiency and avoid the waste of energy sources. Such monitoring and adjustment is critical to ensure proper operation and extend the useful life of the dc governor.

Because when the energy-saving medium-power direct current speed regulating device is abnormal, the energy-saving medium-power direct current speed regulating device can have relatively consistent influence on various performance data, certain correlation exists between different performance data, and the energy-saving medium-power direct current speed regulating device often needs to comprehensively analyze various performance data when dynamically regulating. Therefore, in this embodiment of the present invention, in order to improve the accuracy of the subsequent anomaly monitoring, various performance time series data in the running process of the energy-saving medium-power direct current speed regulating device under the same period are obtained, and in this embodiment of the present invention, the performance parameters may include: the output current, output voltage, output power and load of the energy-saving medium-power direct-current speed regulating device, as well as the motor rotating speed, motor temperature and the like, so that the running condition of the energy-saving medium-power direct-current speed regulating device can be comprehensively reflected.

The specific data acquisition method can be as follows: the output end of the energy-saving medium-power direct-current speed regulating device is measured by using an ammeter and a voltmeter, so that output current and output voltage are collected, and the output power can be obtained by using a formulaTo calculate, among other things,The power is represented by a value representing the power,The voltage is represented by a voltage value,Representing the current; the motor speed and motor temperature can be obtained by using sensors disposed at corresponding positions. The period of all performance time series data acquisition was set to 20 minutes and the sampling time interval was set to 0.1s. In this embodiment of the present invention, since correlation analysis is required for different performance time series data, in order to eliminate the influence of dimension, all the collected data are standardized, a time series curve is drawn for each standardized performance data, the horizontal axis is time, and the vertical axis is a standardized value, so that performance time series data corresponding to each performance parameter is obtained, and at this time, each time corresponds to one value in each performance parameter time series data.

It should be noted that the specific performance parameters are not limited herein; the period of collection and the sampling time interval can be adjusted according to the implementation scene, and are not limited herein.

The data change characteristics of each data point in each performance time sequence data can represent the dynamic change of the running state of the power direct current speed regulating device in the energy-saving mode, and a data basis is provided for subsequent abnormal monitoring, so that the data change characteristics of each data point in each performance time sequence data are obtained simultaneously when the performance time sequence is obtained.

Preferably, the analysis process of the data change characteristics in one embodiment of the present invention includes:

referring now to FIG. 2, a flow chart illustrating a process for analyzing characteristics of data changes in one embodiment of the invention is shown, the process comprising the steps of:

step S201: and merging the numerical characteristics of the data points corresponding to each moment under each performance time sequence data, and determining the performance characteristic value corresponding to each moment.

The performance characteristic value is calculated by combining the numerical characteristics of each moment under various performance time sequence data, and the performance characteristic value integrates the numerical values in all the performance time sequence data at the same moment, so that the overall performance state of the energy-saving medium-power direct current speed regulating device at each moment can be more comprehensively reflected.

For any moment, taking the average value of the values of the data points corresponding to the moment under all performance time sequence data as the performance average value corresponding to the moment, taking the performance average value as the performance characteristic value corresponding to the moment, and recording asAt this timeThe system comprises multidimensional information and characterizes the running state of the system under all performance time sequence data at a certain moment.

Step S202: and selecting one kind of performance time sequence data as the data to be tested, and determining the data change characteristic value of each data point in the data to be tested based on the fluctuation condition of the numerical values of the data points in the data to be tested and combining the difference condition of the performance characteristic values between the corresponding moments of the data points.

The data change characteristic of the data can be represented by the fluctuation condition of the data, so that the numerical fluctuation condition of the data points in each performance time sequence data is selected and analyzed, the data change characteristic value of the data points in each performance time sequence data is determined by combining the performance characteristic value difference between the corresponding moments of the data points, and the data change characteristic value can represent the change characteristic of the data. The method can highlight potential abnormal fluctuation and variation trend in each performance time sequence data, so that the abnormal situation can be accurately identified in the subsequent process.

Since there may be a certain correlation between different performance timing data, for example: the current changes, the motor rotation speed also changes, but the current changes, the load changes or the voltage changes, and the corresponding relevance can be that the load changes or the voltage changes cause the motor rotation speed to change. When the load increases, the corresponding power increases, and there is a significant increase in the heating. Therefore, the numerical fluctuation condition of the data points in each performance time sequence data can be analyzed, the dynamic change condition of each data point in each performance time sequence data can be obtained, and the dynamic change condition is used for measuring the data change characteristic value of each data point; since the performance characteristic values at each time reveal the overall performance among the performance time series data, the difference of the performance characteristic values at each time is also used as an index for measuring the data change characteristic values of the data points in each performance time series data. Thereby providing for the subsequent analysis of the correlation between performance time series data.

Since the fluctuation condition of the data can be represented by the variation trend of the data and the fluctuation of the variation trend, the method for acquiring the variation characteristic value of the data in the embodiment of the invention comprises the following steps:

referring to fig. 3, a method flowchart of a method for obtaining a data change feature value according to an embodiment of the invention is shown, and the method includes the following steps:

Step S301: and in the data to be measured, determining the trend characteristic value of each data point in the data to be measured based on the difference condition of the performance characteristic values between the corresponding moments of the data points.

Taking two moments with the nearest moment distance corresponding to the data point as undetermined moments; and determining the trend characteristic value corresponding to the data point based on the difference between the performance characteristic value of the time corresponding to the data point and the performance characteristic values of the two undetermined times.

Data to be tested is time-series data with performanceFor example, the formula model of the trending feature value includes:

；

Wherein, Representing data to be measuredMiddle (f)Trend eigenvalues of the individual data points; Representing data to be measured Middle (f)Performance characteristic values at times corresponding to the data points; Representing data to be measured Middle (f)Performance characteristic values of undetermined time 1 corresponding to data points; Representing data to be measured Middle (f)Performance characteristic value of undetermined time 2 corresponding to data point.

In the formula model of the trend characteristic value, the difference of the performance characteristic value between the moment corresponding to each data point and the moment nearest in time sequence is analyzed on the whole to obtain the trend characteristic valueThe trend characteristic value can characterize the data change trend at a certain moment from the multidimensional data.

Step S302: in the data to be measured, a fluctuation factor of each data point is determined based on the numerical value change degree among the data points and the fluctuation condition of the data change degree.

In this embodiment of the present invention, the numerical slope is used to reflect the degree of change of the numerical value, so that in the data to be measured, based on the data change trend of the data points, the first fluctuation parameter of each data point is obtained:

For any one data point in the data to be measured, acquiring a data curve of the data to be measured, taking the numerical slope at the data point as a first fluctuation parameter, and recording as . It should be noted that, the slope calculating method is that the difference between the last value and the previous value is compared with the time interval between the two values, and the slope of the last data point in time sequence can be set to be equal to the slope of the last data point.

Then consider the fluctuation of the value slope at the data point, i.e. the fluctuation of the first fluctuation parameter: determining a preset window of the data point based on the time sequence characteristics, and taking the variance of the numerical slopes of all the data points as a second fluctuation parameter of the data point in the preset window corresponding to the data point asThe larger the second fluctuation parameter is, the more severe the fluctuation condition of the numerical value change trend is in a preset window corresponding to the data point.

Finally, taking the product of the first fluctuation parameter and the second fluctuation parameter at the data point as the fluctuation factor at the data point. Data to be tested is time-series data with performanceFor example, the formula model of the fluctuation factor includes:

；

Wherein, Representing data to be measuredMiddle (f)A fluctuation factor of the data points; Representing data to be measured Middle (f)A first fluctuation parameter for the data point; Representing data to be measured Middle (f)A second fluctuation parameter for the data point.

In a formula model of the fluctuation factor, the numerical slope at each data point is used as a first fluctuation parameter to reflect the fluctuation characteristic of the data point, a preset window is constructed, the variance of the numerical slope at all the data points in the window is analyzed and used as a second fluctuation parameter, and finally the first fluctuation parameter and the second fluctuation parameter of each data point in the data to be detected are combined, and the product of the first fluctuation parameter and the second fluctuation parameter is used as the fluctuation factor of each data point in the data to be detected.

It should be noted that, the preset window of each data point is composed of four data points with the nearest time sequence, and the size of the preset window can be adjusted according to the implementation scenario, which is not limited herein.

Because the product of the first fluctuation parameter and the second fluctuation parameter is used as a fluctuation factor, the first fluctuation parameter and the second fluctuation parameter are positively correlated with the fluctuation factor, and the positive correlation can be concretely a multiplication relation, an addition relation, idempotent of an exponential function; therefore, in other embodiments of the present invention, the sum of the first fluctuation parameter and the second fluctuation parameter may be used as the fluctuation factor.

Step S303: and obtaining the data change characteristic value of each data point in the data to be measured according to the trend characteristic value of each data point and the fluctuation factor.

And for any data point in the data to be measured, obtaining a data change characteristic value of the data point under the data to be measured according to the trend characteristic value and the fluctuation factor of the data point, wherein the trend characteristic value and the fluctuation factor are positively correlated with the data change characteristic value. Data to be tested is time-series data with performanceFor example, the formula model of the data change feature value includes:

；

Wherein, Representing data to be measuredMiddle (f)Data change characteristic values of the data points; Representing data to be measured Middle (f)Trend eigenvalues of the individual data points; Representing data to be measured Middle (f)A fluctuation factor of the data points.

In the formula model of the data change characteristic value, in step S301, a trend characteristic value of each data point in the data to be measured is calculated, where the trend characteristic value can reflect the change condition of the performance characteristic value in all performance time series data at the moment corresponding to each data point; the fluctuation factor at each data point calculated in step S302 may characterize the fluctuation of the numerical variation trend of each data point within a certain range. And combining the trend characteristic value corresponding to each data point with the fluctuation factor to obtain a data change characteristic value of each data point, wherein the data change characteristic value is used for representing the data change characteristic at each data point.

Because the trend characteristic value and the fluctuation factor are positively correlated with the data change characteristic value, the positive correlation relationship indicates that the dependent variable increases along with the increase of the independent variable, the dependent variable decreases along with the decrease of the independent variable, and the specific relationship can be multiplication relationship, addition relationship, power of an exponential function and the like; therefore, in other embodiments of the present invention, the sum of the trend characteristic value and the fluctuation factor may be used as the data change characteristic value. Because the fluctuation condition of the data can be directly characterized by the discrete condition of the numerical value, in other embodiments of the present invention, the method for obtaining the characteristic value of the data change can also include:

Selecting one performance time sequence data as to-be-measured data, wherein in the to-be-measured data, for any one data point, two moments with the nearest moment distance corresponding to the data point are used as to-be-measured moments; and determining a trend characteristic value corresponding to the data point based on the difference between the performance characteristic value at the moment corresponding to the data point and the performance characteristic values at the two undetermined moments, wherein the calculation process of the trend characteristic value is the same as that of the trend characteristic value in the step S301.

And determining a preset window of the data point based on the time sequence characteristics, and taking the variance of the numerical values of all the data points as a fluctuation factor of the data point in the preset window corresponding to the data point.

And finally, taking the product of the fluctuation factor and the trend characteristic value of the data point as the data change characteristic value of the data point.

Step S2: in a preset neighborhood corresponding to each data point, determining an association weight corresponding to each data point in each performance time sequence data based on the difference condition of the data change characteristics of the data points under different performance time sequence data; and under any two different performance time sequence data, analyzing the difference of the data change characteristics of all the data points, and combining the association weights of the data points to obtain the association characteristics between any two different performance time sequence data.

Through determining a preset neighborhood corresponding to each data point and analyzing the difference of the data change characteristic values of the data points under different performance time sequence data in the neighborhood, the association relation between the performance time sequence data can be accurately identified, and therefore association weights corresponding to each data point in each performance time sequence data are determined. This approach takes into account the local nature of the data points and the pattern of changes within the neighborhood, helping to capture finer correlation information. After the association weight corresponding to each data point is determined, the association weights of the data points are combined, the difference of the data change characteristic values of all the data points is analyzed, the association degree between time series data of different performances is quantified, the association characteristics are analyzed, and a reference is provided for subsequent abnormal monitoring.

Preferably, in one embodiment of the present invention, the method for acquiring the association weight includes:

Since there may be a possibility of relative delay or relative advance in the variation between the performance time series data, when analyzing the associated weight of each data point, a local range is selected to be set, and the associated weight of each data point is determined by analyzing the corresponding data point under different performance time series data at the moment in the local range.

For convenience of explanation and explanation, a data point is selected as a to-be-measured point in the to-be-measured data, and a preset neighborhood corresponding to the to-be-measured point is determined. Taking the time corresponding to the point to be measured as the time to be measured, and taking the time of all data points in the preset neighborhood corresponding to the point to be measured as the reference time, wherein the reference time comprises the time to be measured.

Taking other performance time sequence data except the data to be measured as comparison data; for any one comparison data, based on the difference between the data change characteristic values of the data points corresponding to the measurement points and each reference moment in the comparison data, the correlation factors of the measurement points and each reference moment in the comparison data can be obtained. Data to be tested is time-series data with performanceFor example, the formula model of the correlation factor includes:

；

Wherein, Representing data to be measuredMiddle (f)Data point and the firstAt the first reference timeCorrelation factors in the individual comparison data; Representing data to be measured Middle (f)Data change characteristic values of the data points; Representing data to be measured Middle (f)Data point ofAt the first reference timeCorresponding data change characteristic values in the comparison data; Expressed in natural constant An exponential function of the base.

In a formula model of the correlation factor, for a to-be-measured point, calculating the difference condition of the data change characteristic value between the to-be-measured point and the data point in the comparison data at each reference moment to obtainIf the value is smaller, it means that the data change condition among the data points is consistent in different performance time sequence data, so the relevance is higher, and thereforeAnd performing negative correlation mapping and normalization processing to realize logic relationship correction, thereby obtaining the correlation factor of the to-be-measured point.

At this time, under the comparison data, the point to be measured has a correlation factor corresponding to each reference time, so that the correlation condition of the data points under different performance time series data can be more accurately represented, and therefore, under the comparison data, the maximum correlation factor is selected as a representative from all the correlation factors of the point to be measured, and is recorded asRepresenting the data to be measuredMiddle (f)Data points at the firstThe largest correlation factor in the comparison data.

Finally, carrying out averaging treatment on the maximum association factors of the points to be measured under all comparison data to obtain association weights of the points to be measured in the data to be measured, and recording the association weights asRepresenting the data to be measuredMiddle (f)Association weights for data points.

It should be noted that the preset neighborhood is composed of four moments that are closest to the point to be measured in time sequence.

For ease of understanding, the method of obtaining the associated weights is illustrated herein: for example, the number of performance time series data is 3, and is denoted as performance time series data a, b, and c, respectively. Taking the performance time sequence data a as the data to be tested, the performance time sequence data b and c are both comparison data, and for the 4 th data point in the performance time sequence data a, the corresponding reference moments are moments 2,3, 4,5 and 6 respectively. And calculating the difference condition between the data change characteristic value of the 4 th data point in the performance time sequence data a and the data change characteristic value of the data point in the comparison data b at the reference time instant 2, thereby obtaining a correlation factor between the 4 th data point in the performance time sequence data a and the data point in the comparison data b at the reference time instant 2, namely obtaining the correlation factor between the 4 th data point in the performance time sequence data a and the 2 nd data point in the comparison data b. Similarly, the correlation factor between the 4 th data point in the performance time series data a and the reference time 3 in the comparison data b can be obtained, and the like, and finally, the correlation factor between the 4 th data point in the performance time series data a and the 5 data points in the comparison data b can be obtained, and the largest correlation factor is selected from the correlation factors. Then, the same process is analyzed in the performance time series data a and the comparison data c, so that the correlation factor between the 4 th data point in the performance time series data a and the 5 th data points in the comparison data c can be obtained, and the largest correlation factor is selected from the correlation factors. At this time, the 4 th data point in the performance time series data a has two maximum association factors, and then the average value of the two maximum association factors is taken as the association weight of the 4 th data point in the performance time series data a.

By the aid of the method, the association weight of each data point in each performance time sequence data can be obtained from the local data points, and then association features among different performance time sequence data can be analyzed from the whole layer.

Preferably, in one embodiment of the present invention, the analysis process of the associated feature includes:

Firstly, combining any two performance time sequence data to obtain all non-repeated data combinations. In each data combination, one of the performance timing data is taken as first data, and the other performance timing data is taken as second data.

Optionally, taking one moment as a target moment, fusing the difference condition of the data change characteristic value of the corresponding data point in the first data at the target moment and the data change characteristic value of the corresponding data point in the second data at the target moment with the association weight of the corresponding data point in the first data and the second data at the target moment, and determining the association characteristic factors of the first data and the second data at the target moment:

And taking the difference between the data change characteristic value of the corresponding data point in the first data at the target moment and the data change characteristic value of the corresponding data point in the second data at the target moment as a difference factor, and taking the value obtained by carrying out negative correlation mapping and normalization on the difference factor as a first adjustment factor of the first data and the second data at the target moment. And taking the average value of the association weight of the corresponding data point in the first data at the target moment and the association weight of the corresponding data point in the second data at the target moment as a second adjustment factor.

And taking the product of the first adjustment factor and the second adjustment factor as an associated characteristic factor of the first data and the second data at the target time. The formula model associated with the feature factor may specifically be, for example:

；

Wherein, Represent the firstThe two performance time sequence data in the data combination are in the firstAssociated feature factors at each moment; Represent the first At a first time instant at first dataAssociated weights for corresponding data points in the set; Represent the first At a second time instantAssociated weights for corresponding data points in the set; Represent the first At a first time instant at first dataData change characteristic values of corresponding data points; Represent the first At a second time instantData change characteristic values of corresponding data points; Expressed in natural constant An exponential function of the base.

In the formula model of the associated feature factors, the difference of the data change feature values of the data points in different performance time series data at the same time is analyzed to obtain the difference factorsThe smaller the value, the more consistent the data change characteristics of the data points in the time sequence data of different performances are, so that the stronger the relevance of the time sequence data of the two performances is, the more negative correlation mapping and normalization are carried out on the difference factors, the logic relationship correction is realized, and the first adjustment factor is obtained. Meanwhile, because each data point in each performance time sequence data has an associated weight, and the greater the associated weight, the stronger the association between the data point and the data points at the same time in all other performance time sequence data is, the average of the associated weights of the data points at the same time in the two performance time sequence data in the data combination is calculated as a second adjustment factor. And finally, taking the product of the first adjustment factor and the second adjustment factor as a correlation characteristic factor between the two performance time sequence data at the same time, wherein the larger the value is, the stronger the correlation between the two performance time sequence data at the same time is.

At this time, in the first data and the second data, each moment corresponds to an associated feature factor, so that the associated feature factors of the first data and the second data at all moments are averaged to obtain associated feature values between the first data and the second data in each data combination, which are recorded asThe associated feature value characterizes the associated feature between the data.

Step S3: determining the nearest neighbor distance corresponding to each moment based on the association characteristics among the time sequence data of different performances and the data change characteristics of the data points of different moments under the time sequence data of different performances; based on the nearest neighbor distance at each moment, the COF algorithm is combined to monitor the abnormal operation of the energy-saving medium-power direct current speed regulating device.

When performing anomaly detection using the COF algorithm, points in the time series data are typically compared with points that are close in time or space, thereby capturing a local variation pattern of the data points and identifying anomalous data points that do not conform to the normal variation pattern. According to the method, the correlation condition between the performance time sequence data is obtained, and the calculation mode of the nearest neighbor distance is redetermined by comprehensively considering the correlation characteristic between different performance time sequence data and the change characteristic of the data point, so that the change characteristic of each data point in the time sequence data can be more accurately depicted, the change mode of the data point is more comprehensively known, the capability of detecting the abnormality of the COF algorithm is enhanced, and the possibility of misjudgment is reduced.

Since the data acquired in this embodiment of the present invention is multidimensional data, the timing at which an abnormality occurs can be determined at the time of abnormality detection. And when calculating the nearest neighbor distance of the moment corresponding to the data point, the method can analyze the whole data, and analyze the similar fluctuation condition of the obtained data change in all the data and the fluctuation characteristic of the corresponding data in the local so as to quantify the nearest neighbor distance. For example: when accessing a load, the data change exhibits a sudden change in local, but if there are multiple similarly distributed data in the overall change of data, it may not be abnormal data that may be a data change caused by normal operation. Meanwhile, the data does not generate mutation but the corresponding data change does not accord with the change trend in all dimension data, and the data can be abnormal data. Therefore, the calculation mode of the nearest neighbor distance at each moment is quantified by analyzing the change consistency of the data in different dimensions and the association characteristic values among the data in different dimensions.

Preferably, in one embodiment of the present invention, the method for acquiring the nearest neighbor distance includes:

optionally, one time is taken as a target time, and other times except the target time are taken as comparison times. The embodiment of the invention analyzes the consistent change condition of the data in the multi-dimensional data and is used for quantifying the nearest neighbor distance of the data, but the data change caused by abnormality can also have consistent performance in the multi-dimensional data, so that in order to avoid the abnormal data to have consistent data characteristics in different performance time sequence data, a preset number of time points adjacent in time sequence of a target time point are taken as the time points to be analyzed, and a chain structure is constructed, and in the embodiment of the invention, the preset number is set to be 2, namely, the nearest neighbor distance of the target time point is determined by analyzing the three time points connected in time sequence.

And then, carrying out change consistency analysis on the difference of the data change characteristic values, the numerical characteristic and the association characteristic value between the two performance time sequence data in the data combination at the target moment and all the comparison moments in different data combinations to obtain a distance factor at the target moment.

And carrying out change consistency analysis on the difference of the characteristic values of the data change in different data combinations at the moment to be analyzed and all the moments which follow in time sequence, the numerical characteristic and the associated characteristic value between two performance time sequence data in the data combination to obtain a distance factor of the moment to be analyzed.

The reason for carrying out the change consistency analysis is that the data points with consistent change conditions can be found out from a plurality of data, so that the data points with special changes generated by normal operation are prevented from being misjudged to be abnormal in the subsequent process.

Finally, taking the sum of the distance factor of the target moment and the distance factors of the two corresponding moments to be analyzed as the nearest neighbor distance of the target moment, and recording as。

It should be noted that the preset number of values may be adjusted according to the implementation scenario, which is not limited herein.

Preferably, in one embodiment of the present invention, the process of obtaining the distance factor through the variation consistency analysis includes:

Determining a first change parameter of the target time and each comparison time under the first data in each data combination based on the difference of the data change characteristic values of the target time and each comparison time and the numerical value difference; determining a second change parameter of the target time and each comparison time under the second data based on the difference of the data change characteristic values of the target time and each comparison time in the second data and the numerical value difference; taking the difference between the first variation parameter and the second variation parameter as a variation parameter difference value; and taking the product of the value obtained after carrying out negative correlation mapping and normalization on the variation parameter difference value and the correlation characteristic value as a variation consistency factor. Normalizing the sum of the change consistency factors of the target time and each comparison time under all data combinations to obtain distance parameters, wherein the target time and each comparison time have one distance parameter, and taking the maximum value of the distance parameters of the target time and all comparison times as the distance factor of the target time to be recorded as 。

Time of day of the target time of dayFor example, the formula model for varying the consistency factor includes:

；

Wherein, Is shown in the firstIn the data combination, the target timeAnd the firstA variation consistency factor between the individual comparison moments; Represent the first Correlation characteristic values between two performance time sequence data in the data combination; Is shown in the first In the data combination, the target timeAt the first dataData change characteristic values of corresponding data points; Is shown in the first In the data combination, the target timeAt the first dataCorresponding to the number of the corresponding number; Is shown in the first In the data combination, the target timeAt the second dataData change characteristic values of corresponding data points; Is shown in the first In the data combination, the target timeAt the second dataCorresponding to the number of the corresponding number; Is shown in the first In the data combination, the target timeIs the first of (2)At the first data at the moment of comparisonData change characteristic values of corresponding data points; Is shown in the first In the data combination, the target timeIs the first of (2)At the first data at the moment of comparisonCorresponding to the number of the corresponding number; Is shown in the first In the data combination, the target timeIs the first of (2)At the second dataData change characteristic values of corresponding data points; Is shown in the first In the data combination, the target timeIs the first of (2)At the second dataCorresponding to the numerical value of the corresponding code.

In the formula model of the change consistency factor, the larger the correlation characteristic value of the two performance time sequence data in the data combination is, the stronger the correlation between the two performance time sequence data is, the more consistent the data change condition is possible, so the correlation characteristic value is used as an index for calculating the change consistency factor. Then calculating the change condition of the target time and each comparison time in the first data in the data combination to obtain a first change parameterSimilarly, the change condition of the target time and each comparison time in the second data in the data combination is calculated to obtain a second change parameterIf the first variation parameter is relatively close to the second variation parameter, i.e. the variation parameters differThe smaller the data point corresponding to the target moment and the comparison moment is, the more consistent the change trend or characteristic is shown in different performance time sequence data, the change mode is shown to exist in the data, so that the change parameter difference value is subjected to negative correlation mapping and normalization, the combination of the value after logic relation correction and the association characteristic value between the two performance time sequence data in the data combination is realized, and the change consistency factor between the target moment and the comparison moment is obtained.

The formula model of the distance parameter comprises:

；

Wherein, Indicating the target timeAnd the firstDistance parameters between the individual comparison moments; Is shown in the first In the data combination, the target timeAnd the firstA variation consistency factor between the individual comparison moments; representing a total number of data combinations; representing the normalization function.

In a formula model of the distance parameter, the change consistency factors of the target time and the comparison time under all data combinations can be comprehensively analyzed, so that the distance parameter for measuring the distance between the target time and the comparison time is obtained, and the greater the change consistency of the target time and the comparison time under more data combinations is, the greater the distance parameter is, so that the nearest neighbor distance of the target time calculated in the follow-up process is increased, and the recognition of the target time as an abnormality is avoided in the follow-up abnormality detection process.

For ease of understanding, a specific calculation procedure of the distance factor at the target time is illustrated here: for example, the number of performance time series data is 3, respectively denoted as performance time series data a, b, c, and the length of each performance time series data is 6, that is, has 6 time instants, and the data combination is (a, b), (a, c), (b, c). Taking time 1 as the target time, then times 2,3, 4, 5, 6 are all comparison times. In the data combination (a, b), calculating a change consistency factor between the target time 1 and the comparison time 2, in the data combination (a, c), calculating a change consistency factor between the target time 1 and the comparison time 2, in the data combination (b, c), calculating a change consistency factor between the target time 1 and the comparison time 2, wherein three change consistency factors exist between the target time 1 and the comparison time 2, and normalizing the sum of the three change consistency factors to obtain a distance parameter between the target time 1 and the comparison time 2. Then, in the same way, in the data combination (a, b), the change consistency factor between the target time 1 and the contrast time 3 is calculated, in the data combination (a, c), the change consistency factor between the target time 1 and the contrast time 3 is calculated, in the data combination (b, c), the change consistency factor between the target time 1 and the contrast time 3 is calculated, at this time, three change consistency factors are provided between the target time 1 and the contrast time 3, the sum of the three change consistency factors is normalized, so as to obtain the distance parameter … between the target time 1 and the contrast time 3, and so on, the distance parameters between the target time 1 and all the contrast times can be obtained, namely, the target time 1 has 6 distance parameters, and the largest distance parameter is taken as the distance factor of the target time 1.

The distance factor is obtained for each time to be analyzed in the same way: for any moment to be analyzed, taking all moments after the moment to be analyzed in time sequence as reference moments; determining a first change parameter corresponding to the time to be analyzed and each reference time in the first data based on the difference and the numerical value difference of the characteristic value of the data change in the first data between the time to be analyzed and each reference time; determining a second change parameter corresponding to the time to be analyzed and each reference time in the second data based on the difference and the numerical value difference of the characteristic value of the data change in the second data of the time to be analyzed and each reference time; taking the difference between the first variation parameter and the second variation parameter as a variation parameter difference value; the product of the value obtained after carrying out negative correlation mapping and normalization on the variation parameter difference value and the correlation characteristic value is used as a variation consistency factor; normalizing the sum of the change consistency factors of the moment to be analyzed and each reference moment under all data combinations to obtain a distance parameter; and taking the maximum value of the distance parameters of the moment to be analyzed and all the reference moments as a distance factor of the moment to be analyzed.

So far, the nearest neighbor distance of each moment can be obtained through the process. In the process of calculating the nearest neighbor distance of each time, the embodiment of the invention adopts a chain structure, namely, the last two times adjacent in time sequence form a chain, so that the nearest neighbor distance of each time is determined by analyzing the data change characteristics of the time on the chain in different performance time sequence data, but the nearest neighbor distance of the last time and the last-last time in time sequence cannot form the chain structure, so that the nearest neighbor distance of the last time and the last-time can be set to be consistent with the nearest neighbor distance of the last-third time. In other embodiments of the present invention, the process of obtaining the nearest neighbor distance at the last time and the last but last time may also be: the distance factor of the device is directly used as the nearest neighbor distance.

In the embodiment of the invention, the mode of determining the nearest neighbor distance of each moment in the COF algorithm is changed, so that the nearest neighbor distance of each moment at the moment synthesizes the information of multi-dimensional data, and the local and the whole characteristics are analyzed, so that the COF abnormality score of each moment can be more accurately calculated continuously based on the COF algorithm. It should be noted that, the COF anomaly score obtained at each moment based on the COF algorithm is a technical means well known to those skilled in the art, and will not be described herein.

Finally, the anomaly time may be determined based on the COF anomaly score at each time. The method for acquiring the abnormal time comprises the following steps:

And taking the moment when the COF abnormality score is larger than a preset abnormality threshold value as the abnormality moment. It should be noted that the preset anomaly threshold is set to 2, and the specific numerical value implementer can adjust according to the implementation scenario, which is not limited herein.

In summary, the embodiment of the invention firstly obtains various performance time sequence data in the operation process of the energy-saving medium-power direct current speed regulating device under the same time period, and provides a data base for subsequent anomaly monitoring. And then, for each moment, analyzing the numerical characteristics of the energy-saving medium-power direct-current speed regulating device under different performance time sequence data so as to determine a performance characteristic value corresponding to each moment, wherein the performance characteristic value can more comprehensively reflect the performance of the energy-saving medium-power direct-current speed regulating device under each moment. Because abnormal fluctuation and abnormal change conditions are usually generated in the data points at abnormal moments, in each performance time sequence data, the numerical fluctuation conditions of the data points are analyzed, and the characteristic value of the data change at each data point in each performance time sequence data is determined by combining the characteristic value of the performance at each moment, so that the characteristic value of the change at each data point is used for representing the change characteristic at the data point, and further, the dynamic change of the running state of the power direct current speed regulating device in the energy saving mode is better reflected. Further, the change condition of the data in the local range is analyzed, so that a preset neighborhood corresponding to each data point is determined, the difference of the data change characteristic values among the data points under different performance time sequence data is analyzed in the neighborhood, the association weight of the data points is determined, then the difference of the change condition of the data points in the different performance time sequence data is combined, the association characteristic values among the different performance time sequence data are obtained, and the association characteristic values represent the relativity among the different performance time sequence data. Then, through the association characteristic values among different performance time sequence data and the data change characteristic values of the data points under different performance time sequence data, the nearest neighbor distance corresponding to each moment is determined, and the method for determining the nearest neighbor distance not only considers the local relation among the data points, but also combines the relativity among different performance time sequence data, so that the moment when the power direct current speed regulating device in the energy-saving type is abnormal can be detected more accurately in the subsequent process, the subsequent COF abnormal score is calculated based on a COF algorithm, the moment when the abnormality occurs can be screened more accurately, and the accuracy of abnormality monitoring is improved.

The embodiment also provides an operation state monitoring system for the energy-saving type medium-power direct current speed regulating device, which comprises a processor, a memory and a computer program, wherein the memory is used for storing the corresponding computer program, the processor is used for running the corresponding computer program, and the computer program can realize the steps of the operation state monitoring method of any one of the energy-saving type medium-power direct current speed regulating devices when running on the processor.

Referring to fig. 4, a schematic system structure diagram of an operation state monitoring system for an energy-saving medium-power dc speed regulating device according to an embodiment of the invention is shown, including a processor 400, a memory 401, a bus 402 and a communication interface 403, where the processor 400, the communication interface 403 and the memory 401 are connected by the bus 402; where memory 401 may include a high-speed random access memory, bus 402 may be an ISA bus, a PCI bus, an EISA bus, or the like, and processor 400 may be an integrated circuit chip having signal processing capabilities.

Referring to fig. 5, a system block diagram of an operation state monitoring system for an energy-saving medium-power dc speed regulating device according to an embodiment of the invention is shown, including: data acquisition and change feature analysis module 501: step S1 in the method is implemented, where the multidimensional data relevance analysis module 502: for implementing step S2 in the above method, the anomaly monitoring module 503: for implementing step S3 in the above method.

An embodiment of the present invention further provides a computer readable storage medium corresponding to the method provided in the foregoing embodiment, referring to fig. 6, the storage medium is shown as an optical disc, and a computer program (i.e. a program product) is stored on the storage medium, where the computer program, when executed by a processor, performs the method provided in any of the foregoing embodiments.

It should be noted that, examples of the computer readable storage medium may also include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), read Only Memory (ROM), and other optical and magnetic storage media, which are not described herein in detail.

It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments.

Claims

1. An operating state monitoring method for an energy-saving medium-power direct current speed regulating device, which is characterized by comprising the following steps:

Determining the nearest neighbor distance corresponding to each moment based on the association characteristics among the time sequence data of different performances and the data change characteristics of the data points of different moments under the time sequence data of different performances; based on the nearest neighbor distance at each moment, the COF algorithm is combined to monitor the operation abnormality of the energy-saving medium-power direct current speed regulating device;

The acquiring the data change characteristic of each data point in each performance time sequence data comprises the following steps:

fusing the numerical characteristics of the data points corresponding to each moment under each performance time sequence data, and determining the performance characteristic value corresponding to each moment; selecting one type of performance time sequence data as data to be measured, and determining a data change characteristic value of each data point in the data to be measured based on the fluctuation condition of the values of the data points in the data to be measured and combining the difference condition of the performance characteristic values between the corresponding moments of the data points;

Under any two different performance time sequence data, analyzing the difference of the data change characteristics of all data points, and combining the association weights of the data points to obtain the association characteristics between any two different performance time sequence data, wherein the method comprises the following steps:

Averaging the correlation characteristic value factors of the first data and the second data at all moments to obtain the correlation characteristic value between the first data and the second data;

the nearest neighbor distance acquisition method comprises the following steps:

taking the sum of the distance factor of the target moment and the distance factor of all corresponding moments to be analyzed as the nearest neighbor distance of the target moment;

The process of obtaining the distance factor through the change consistency analysis comprises the following steps:

2. The method for monitoring the operation state of the energy-saving medium-power direct current speed regulating device according to claim 1, wherein the plurality of performance time sequence data at least comprises: current timing data, voltage timing data, motor speed timing data, and load timing data.

3. The method for monitoring the operation state of the energy-saving medium-power direct current speed regulating device according to claim 1, wherein the method for acquiring the characteristic value of the data change comprises the following steps:

In the data to be measured, for any one data point;

4. The method for monitoring the operation state of the energy-saving medium-power direct current speed regulating device according to claim 1, wherein the method for acquiring the association weight comprises the following steps:

5. The method for monitoring the operation state of the energy-saving medium-power direct current speed regulating device according to claim 1, wherein the method for acquiring the correlation characteristic factor comprises the following steps:

6. The method for monitoring the operation state of the energy-saving medium-power direct current speed regulating device according to claim 1, wherein the method for acquiring the change consistency factor comprises the following steps: