CN114330515A - Bridge monitoring data abnormity diagnosis and repair method - Google Patents
Bridge monitoring data abnormity diagnosis and repair method Download PDFInfo
- Publication number
- CN114330515A CN114330515A CN202111533049.XA CN202111533049A CN114330515A CN 114330515 A CN114330515 A CN 114330515A CN 202111533049 A CN202111533049 A CN 202111533049A CN 114330515 A CN114330515 A CN 114330515A
- Authority
- CN
- China
- Prior art keywords
- data
- abnormal
- sensor
- network model
- sensors
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000012544 monitoring process Methods 0.000 title claims abstract description 20
- 230000008439 repair process Effects 0.000 title claims abstract description 20
- 238000003745 diagnosis Methods 0.000 title claims abstract description 14
- 230000002159 abnormal effect Effects 0.000 claims abstract description 65
- 238000003062 neural network model Methods 0.000 claims abstract description 22
- 238000012549 training Methods 0.000 claims abstract description 15
- 239000013598 vector Substances 0.000 claims abstract description 13
- 230000005856 abnormality Effects 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 230000002457 bidirectional effect Effects 0.000 claims description 26
- 230000015654 memory Effects 0.000 claims description 17
- 230000006870 function Effects 0.000 claims description 16
- 238000010606 normalization Methods 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 10
- 238000009826 distribution Methods 0.000 claims description 6
- 230000007787 long-term memory Effects 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 230000006403 short-term memory Effects 0.000 claims description 3
- 239000006185 dispersion Substances 0.000 claims description 2
- 238000000605 extraction Methods 0.000 abstract description 3
- 238000009412 basement excavation Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000013145 classification model Methods 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 2
- 230000009191 jumping Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 230000002277 temperature effect Effects 0.000 description 1
Images
Landscapes
- Testing And Monitoring For Control Systems (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a bridge monitoring data abnormity diagnosis and repair method which comprises the steps of firstly, carrying out data acquisition and preprocessing on a multi-point sensor, normalizing data and eliminating interference of an amplitude absolute value. Then carrying out abnormal feature extraction of 11-dimensional feature vectors on the normalized data; and then constructing a BilSTM neural network model, training a network, storing the network when the prediction accuracy is more than or equal to 95 percent for the subsequent abnormal identification and positioning of real-time data, inputting the real-time data into the model to judge and position abnormal data segments, and finishing the data abnormality judgment of the first stage. And finally, generating training data of the countermeasure network model according to the data with proper length before the abnormity as a condition, training the acquired multi-sensor correlation model according to the model, taking the normal sensor data in the abnormal data period as input, predicting the data in the abnormal sensor period as filling replacement of the abnormal data period, and thus realizing the full-process automatic identification and repair of the abnormal trend of the bridge monitoring data.
Description
Technical Field
The invention relates to a bridge monitoring data abnormity diagnosis and repair method, and belongs to the technical field of bridge structure health monitoring data cleaning and repair.
Background
The premise of the function of the bridge monitoring system is to acquire various data which truly reflect the structural performance. However, the data is inevitably adversely affected by various factors such as sensor aging faults, environmental noise interference, transmission packet loss and compression distortion in the processes of data measurement, acquisition, transmission, storage and the like, so that the monitored data is distorted and abnormal in amplitude, distribution and trend. The main types of data anomalies include missing, jumping, noise, drift, and trending anomalies. Wherein, the first four can carry out effective identification by comparing the time-lapse data of the sensor. However, for the abnormal trend, the data is often interfered by long periods, and the phenomenon that a large area of the data continuously deviates from the normal value is caused. Because the structure has continuous abnormal temperature effect, the response data has the possibility of continuous deviation, so the abnormal type is difficult to directly distinguish only from the self historical data, and the data usually continuously appears in a large area, and the difficulty of real-time repair only from the autoregressive angle is huge.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: the method comprises the steps of firstly adopting a supervised learning paradigm to realize identification and positioning of abnormal data, establishing a generated confrontation neural network model on the basis of the identification and positioning of the abnormal data to predict and replace an abnormal data section, and completing full-process automatic identification and repair of sensor data abnormity.
The invention adopts the following technical scheme for solving the technical problems:
a bridge monitoring data abnormity diagnosis and repair method comprises the following steps:
step 2, carrying out normalization processing on the preprocessed sensor data to obtain normalized data;
and 3, extracting 11-dimensional feature vectors of the normalized data by taking minutes as a time scale, wherein the 11-dimensional feature vectors comprise: a mean, a median, a standard deviation, a root mean square, a magnitude, a difference value, a binary value, an eighty percent value, a ratio of a maximum value to an eighty percent value, a ratio of a mean to a magnitude, and a ratio of a mean to an eighty percent value;
step 4, constructing a bidirectional long-and-short term memory neural network model, and training the bidirectional long-and-short term memory neural network model by using pre-marked label data to obtain a trained bidirectional long-and-short term memory neural network model; inputting the 11-dimensional characteristic vectors extracted in the step 3 into a trained bidirectional long-time memory neural network model, diagnosing abnormal data, and entering a step 5 when abnormal data exists in diagnosis;
and 5, hollowing data corresponding to the abnormal data time period, filling the data with nan, generating a confrontation network model by taking data of all sensors 3 hours before the abnormal sensor occurs as a data set training condition, acquiring a nonlinear relation between the data before the abnormal sensor occurs and data of other normal sensors in the same time period, generating the confrontation network model by using the data of nan filling the abnormal data time period and the data of other normal sensors in the abnormal time period as input conditions, predicting the data of the abnormal sensor in the abnormal time period, and replacing nan with the predicted data to obtain repaired data.
As a preferred embodiment of the present invention, the critical sections in step 1 include 1/4 section, midspan section, 3/4 section and main tower top section of cable bridge.
As a preferred scheme of the present invention, the preprocessing of the sensor data in step 1 specifically includes: and data acquisition is carried out on the data acquired by each sensor, so that the data length corresponding to each sensor is ensured to be consistent and the data length does not contain nan values.
As a preferable scheme of the present invention, the normalization processing in step 2 adopts a dispersion normalization method, and the specific formula is as follows:
wherein x is(i,j)Denotes x(i,j)Normalized data, x(i,j)Represents the preprocessed data, x, of the sensor numbered j at the ith critical sectioniRepresenting the set of pre-processed data for all sensors.
As a preferable scheme of the present invention, the bidirectional long and short term memory neural network model in step 4 includes a first bidirectional LSTM layer, a second bidirectional LSTM layer, a third bidirectional LSTM layer, a fully-connected layer, and a Softmax layer, which are connected in sequence.
As a preferred scheme of the present invention, the loss function of the bidirectional long-and-short term memory neural network model in step 4 is a cross entropy loss function, which specifically includes:
wherein, los represents a cross entropy loss function; y is a sample label, if the sample belongs to a positive example, namely data is abnormal, the value of the sample is 1, otherwise, the value of the sample is 0 if the data is normal;
is the probability that the model predicted the sample to be a positive case;
when the bidirectional long-time memory neural network model is trained, the model is stored when the classification accuracy rate is more than 95 percent, and the model is used as the trained bidirectional long-time memory neural network model, wherein the classification accuracy rate specifically comprises the following steps:
Accuracy＝(TP+TN)/(ALL)
where Accuracy represents the classification Accuracy, TP represents the number of true positive classifications, TN represents the number of true negative classifications, and ALL represents the total number of labels in the pre-labeled label data.
As a preferred embodiment of the present invention, the condition in step 5 generates a confrontation network model, and the objective function of network optimization is expressed as:
wherein G and D represent the generator and the discriminator, respectively, V represents the objective function, E represents the distribution expectation, f represents the condition vector, t represents the input, p represents the output, anddata(t) denotes true samples, z denotes initial generated noise, pz(z) represents the a priori noise distribution.
Compared with the prior art, the invention adopting the technical scheme has the following technical effects:
1. based on the data correlation of multiple sensors, the invention identifies the abnormal sensors and accurately positions the abnormal data segments in the minute scale, and can realize the accurate identification of the data abnormal types in different time scales, such as jumping point, trend abnormality and the like, by benefiting from the network architecture and the sensitivity characteristic of the BilSTM network to the characteristics in different time lengths. And then generating training data of the countermeasure network model according to data with proper length before abnormality as a condition, training the acquired multi-sensor correlation model according to the model, predicting data of the abnormal sensor in the period by taking normal sensor data in the abnormal data period as input, and performing filling replacement on the abnormal data period, thereby realizing full-process automatic identification and repair of the abnormal trend of the bridge monitoring data.
2. The invention provides a two-stage monitoring data restoration method of firstly identifying and then restoring by adopting a bidirectional LSTM model and a condition generation confrontation network model based on data correlation, which realizes the unification of identification and automatic restoration when data is abnormal and forms a full-flow automatic data cleaning method. And thanks to the given feature extraction method and the relevance restoration method, the method can effectively identify and restore the abnormal type which is traditionally difficult to distinguish only by the historical data of the data.
3. The invention can effectively learn the multi-sensor related mapping relation by constructing and generating the confrontation neural network model for large-area data abnormity, and can well predict the abnormal data section based on the model and other sensor data in abnormal time period, thereby completing the repair of the large-area abnormal condition, and the large-area repaired data is still reliable and accurate.
Drawings
FIG. 1 is a flow chart of a bridge monitoring data anomaly diagnosis and repair method according to the present invention;
FIG. 2 is a BilSTM classification model for anomaly identification according to the present invention;
FIG. 3 is a conditional generative countermeasure network model for data repair according to the present invention;
FIG. 4 is a simulation verification result of the method of the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention.
Because the sensors at different positions of the same bridge structure have obvious correlation in the amplitude and the change rule of data, the method has good theoretical basis for judging data trend abnormity based on the data correlation. Based on data correlation, the long-time and short-time neural memory network is adopted to capture the abnormal features of the data signals in the time frequency domain, so that the technical prospect of rapidly realizing the judgment of a large amount of data abnormity is achieved. The feasibility of the method was also verified by statistical analysis of the large amount of data. Meanwhile, the generated countermeasure neural network can acquire a correlation model of the concerned sensor and other sensors before the abnormality occurs, and the real data of the abnormal sensor in the abnormal data section can be accurately predicted by means of the model and the time period data of the sensors except the concerned sensor after the abnormality occurs. This provides a reliable basis for data anomaly identification and repair of the monitored data.
As shown in fig. 1, the present invention provides a bridge monitoring data anomaly diagnosis and repair method for generating a countermeasure network based on data correlation conditions, which mainly includes the following steps:
(1) and (6) data acquisition and processing. Collecting data of a plurality of sensors at multiple points of a bridge, numbering the data of the sensors, and preprocessing the data collected by the sensors;
the preprocessing work comprises data interception of data collected by the sensors, and the data length of each sensor is consistent and does not contain nan values.
(2) And carrying out normalization processing on all the sensor data. The deviation normalization method is adopted for normalization processing, so that errors caused by absolute values are eliminated, and the specific formula is as follows:
wherein x is(i,j)Denotes x(i,j)Normalized data, x(i,j)Represents the preprocessed data, x, of the sensor numbered j at the ith critical sectioniRepresenting the set of pre-processed data for all sensors.
(3) Extracting features, and calculating 11-dimensional feature vectors by taking each minute as a scale, wherein the method comprises the following steps: carrying out dimension reduction and feature extraction on the original data by using the average value, the median value, the standard deviation, the root mean square, the amplitude (the difference between the maximum value and the minimum value), the difference value, the binary value, the eighty-percent value, the ratio of the maximum value to the eighty-percent value, the ratio of the average value to the amplitude and the ratio of the average value to the eighty-percent value; the 11-dimensional features provided by the invention are key feature parameters which are obtained by repeatedly screening on the basis of 40 statistical feature index libraries and are sensitive to bridge monitoring data abnormity, and have obvious benefits for improving the model accuracy and reducing the training cost;
(4) and (6) positioning abnormal data. As shown in fig. 2, a bidirectional long-and-short term memory neural network model BiLSTM including 3 bidirectional LSTM layers, a full-link layer and a Softmax layer is constructed, and abnormal data is classified. The network architecture is an exploratory trial result of the long and short memory neural network model in a data abnormity classification task, and not only can the accuracy of a training set be met, but also overfitting can be prevented. Performing model training by using the label data, and storing the model when the classification accuracy reaches more than 95% for automatic abnormal recognition of new data;
the loss function of the BilSTM model adopts a cross entropy loss function:
wherein, los represents a cross entropy loss function; y is a sample label, if the sample belongs to a positive example, namely data is abnormal, the value of the sample is 1, otherwise, the value of the sample is 0 if the data is normal;
is the probability that the model predicted the sample to be a positive case;
when a bidirectional long-time memory neural network model is trained, when the classification accuracy rate is more than 95%, the model is stored and used for predicting and classifying new data, and the classification accuracy rate is specifically as follows:
Accuracy＝(TP+TN)/(ALL)
where Accuracy represents the classification Accuracy, TP represents the number of true positive classifications, TN represents the number of true negative classifications, and ALL represents the total number of labels in the pre-labeled label data.
(5) And (5) data repair. The data for the abnormal period was hollowed out and filled in with nan. And (3) generating a confrontation network (CGAN) model by taking all data of the data volume 3 hours before the occurrence of the abnormity as data set training conditions, and acquiring the nonlinear relation of the data of the abnormal sensor before the occurrence of the abnormity and other normal sensors. Filling nan data of the abnormal sensor data time interval, taking the data of the abnormal sensor data time interval and all other normal similar sensor data of the same time interval as model input data, predicting abnormal sensor excavation data, and filling excavation data with the predicted data to obtain repaired data;
in the generator, G, the a priori noise profile P, as shown in FIG. 3z(z) and condition vector y are combined as inputs to the first layer of the neural network, usually against the training framework with considerable flexibility for the composition of this implicit representation; in the discriminator D, the true sample pdataThe input t of (t) and the condition vector f are the inputs to a discriminant function, typically a multi-layer perceptron. The objective function of network optimization can be expressed as:
where G and D denote the generator and the discriminator, respectively, E denotes the distribution expectation, f denotes the condition vector, t denotes the input, and z denotes the initial generated noise.
Examples
The following describes a specific implementation process of the present invention by taking the determination and the repair of the abnormal trend of the GPS data of the bridge across the river as an example.
(1) And (6) data acquisition and processing. Data at key sections of the bridge are collected, including 1/4, midspan, 3/4 and main tower top section positions of the cable bridge, 2 sensors per section, and 8 sensors in total are numbered 1-8.
(2) And (6) normalization processing. And carrying out normalization processing on data of all 8 GPS sensors of the same type in one day, and eliminating errors caused by absolute values of different interface position coordinates.
(3) The basic identification scale is divided by the amount of data per hour. Then, taking the minute as a time scale to extract 11-dimensional characteristic parameters of each sensing datum, wherein the 11-dimensional characteristic parameters comprise: mean, median, standard deviation, root mean square, magnitude (difference between maximum and minimum), difference value, binary, eighty percent value, maximum divided by eighty percent value, mean to magnitude (difference between maximum and minimum), mean to eighty percent value.
(4) And (4) data anomaly identification and classification. And automatically classifying and positioning the data abnormity according to the data abnormity classification model stored by the modeling method. And inputting the characteristic matrix corresponding to each sensor, and identifying the abnormal data sensor and the abnormal data occurrence time of the abnormal data sensor.
(5) And (5) data repair. The data for the abnormal period was hollowed out and filled in with nan. Generating a confrontation network model under all homogeneous data training conditions of data volume 3 hours before an abnormality occurs to acquire a nonlinear relation between data of the abnormal sensor before the abnormality occurs and data of all other sensors, then using nan-filled sensor data and data of all other homogeneous data in the period as model input data to predict abnormal sensor excavation data, and finally filling the excavation data with the predicted data to obtain repaired data, wherein a simulation verification result is shown in figure 4.
The above embodiments are only for illustrating the technical idea of the present invention, and the protection scope of the present invention is not limited thereby, and any modifications made on the basis of the technical scheme according to the technical idea of the present invention fall within the protection scope of the present invention.
Claims (7)
1. A bridge monitoring data abnormity diagnosis and repair method is characterized by comprising the following steps:
step 1, arranging sensors at all key sections of a bridge, numbering the sensors, collecting data of all the sensors, and preprocessing the data of the sensors, wherein the types of the sensors arranged at all the key sections are the same;
step 2, carrying out normalization processing on the preprocessed sensor data to obtain normalized data;
and 3, extracting 11-dimensional feature vectors of the normalized data by taking minutes as a time scale, wherein the 11-dimensional feature vectors comprise: a mean, a median, a standard deviation, a root mean square, a magnitude, a difference value, a binary value, an eighty percent value, a ratio of a maximum value to an eighty percent value, a ratio of a mean to a magnitude, and a ratio of a mean to an eighty percent value;
step 4, constructing a bidirectional long-and-short term memory neural network model, and training the bidirectional long-and-short term memory neural network model by using pre-marked label data to obtain a trained bidirectional long-and-short term memory neural network model; inputting the 11-dimensional characteristic vectors extracted in the step 3 into a trained bidirectional long-time memory neural network model, diagnosing abnormal data, and entering a step 5 when abnormal data exists in diagnosis;
and 5, hollowing data corresponding to the abnormal data time period, filling the data with nan, generating a confrontation network model by taking data of all sensors 3 hours before the abnormal sensor occurs as a data set training condition, acquiring a nonlinear relation between the data before the abnormal sensor occurs and data of other normal sensors in the same time period, generating the confrontation network model by using the data of nan filling the abnormal data time period and the data of other normal sensors in the abnormal time period as input conditions, predicting the data of the abnormal sensor in the abnormal time period, and replacing nan with the predicted data to obtain repaired data.
2. The bridge monitoring data anomaly diagnosis and restoration method according to claim 1, wherein the key sections in step 1 include 1/4 section, midspan section, 3/4 section and main tower top section of a cable bridge.
3. The bridge monitoring data abnormality diagnosis and restoration method according to claim 1, wherein the preprocessing of the sensor data in step 1 is specifically: and data acquisition is carried out on the data acquired by each sensor, so that the data length corresponding to each sensor is ensured to be consistent and the data length does not contain nan values.
4. The method for diagnosing and repairing the abnormality of the bridge monitoring data according to claim 1, wherein the normalization processing in the step 2 is a dispersion normalization method, and a concrete formula is as follows:
wherein x is(i,j)Denotes x(i,j)Normalized data, x(i,j)Represents the preprocessed data, x, of the sensor numbered j at the ith critical sectioniRepresenting the set of pre-processed data for all sensors.
5. The bridge monitoring data anomaly diagnosis and repair method according to claim 1, wherein the bidirectional long-term and short-term memory neural network model in the step 4 comprises a first bidirectional LSTM layer, a second bidirectional LSTM layer, a third bidirectional LSTM layer, a full connection layer and a Softmax layer which are connected in sequence.
6. The bridge monitoring data abnormality diagnosis and restoration method according to claim 1, wherein the loss function of the bidirectional long-term and short-term memory neural network model in step 4 is a cross entropy loss function, and specifically comprises:
wherein, los represents a cross entropy loss function; y is a sample label, if the sample belongs to a positive example, namely data is abnormal, the value of the sample is 1, otherwise, the value of the sample is 0 if the data is normal;
is the probability that the model predicted the sample to be a positive case;
when the bidirectional long-time memory neural network model is trained, the model is stored when the classification accuracy rate is more than 95 percent, and the model is used as the trained bidirectional long-time memory neural network model, wherein the classification accuracy rate specifically comprises the following steps:
Accuracy＝(TP+TN)/(ALL)
where Accuracy represents the classification Accuracy, TP represents the number of true positive classifications, TN represents the number of true negative classifications, and ALL represents the total number of labels in the pre-labeled label data.
7. The method for diagnosing and repairing abnormality of bridge monitoring data according to claim 1, wherein the condition of step 5 is a countermeasure network model, and an objective function of network optimization is expressed as:
wherein G and D represent the generator and the discriminator, respectively, V represents the objective function, E represents the distribution expectation, f represents the condition vector, t represents the input, p represents the output, anddata(t) denotes true samples, z denotes initial generated noise, pz(z) represents the a priori noise distribution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111533049.XA CN114330515A (en) | 2021-12-15 | 2021-12-15 | Bridge monitoring data abnormity diagnosis and repair method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111533049.XA CN114330515A (en) | 2021-12-15 | 2021-12-15 | Bridge monitoring data abnormity diagnosis and repair method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114330515A true CN114330515A (en) | 2022-04-12 |
Family
ID=81052727
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111533049.XA Pending CN114330515A (en) | 2021-12-15 | 2021-12-15 | Bridge monitoring data abnormity diagnosis and repair method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114330515A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118656648A (en) * | 2024-08-20 | 2024-09-17 | 江苏省沿海开发投资有限公司 | Algae cultivation data processing method and system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109583570A (en) * | 2018-11-30 | 2019-04-05 | 重庆大学 | The method for determining bridge health monitoring system abnormal data source based on deep learning |
| CN111695598A (en) * | 2020-05-11 | 2020-09-22 | 东南大学 | Monitoring data abnormity diagnosis method based on generation countermeasure network |
| CN113240011A (en) * | 2021-05-14 | 2021-08-10 | 烟台海颐软件股份有限公司 | Deep learning driven abnormity identification and repair method and intelligent system |
-
2021
- 2021-12-15 CN CN202111533049.XA patent/CN114330515A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109583570A (en) * | 2018-11-30 | 2019-04-05 | 重庆大学 | The method for determining bridge health monitoring system abnormal data source based on deep learning |
| CN111695598A (en) * | 2020-05-11 | 2020-09-22 | 东南大学 | Monitoring data abnormity diagnosis method based on generation countermeasure network |
| CN113240011A (en) * | 2021-05-14 | 2021-08-10 | 烟台海颐软件股份有限公司 | Deep learning driven abnormity identification and repair method and intelligent system |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118656648A (en) * | 2024-08-20 | 2024-09-17 | 江苏省沿海开发投资有限公司 | Algae cultivation data processing method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111737909B (en) | Structural health monitoring data anomaly identification method based on space-time graph convolutional network | |
| CN112785091B (en) | Method for carrying out fault prediction and health management on oil field electric submersible pump | |
| CN111382542A (en) | Road electromechanical equipment life prediction system oriented to full life cycle | |
| CN118154174B (en) | Intelligent operation and maintenance cloud platform for industrial equipment | |
| CN114707234B (en) | Aero-engine residual life prediction method based on Bayesian deep learning | |
| CN117609836A (en) | Electromagnetic sensitivity prediction and health management method for integrated module | |
| CN111611294A (en) | Star sensor data anomaly detection method | |
| CN117636073B (en) | Concrete defect detection method, device and storage medium | |
| CN113988210A (en) | Method and device for restoring distorted data of structure monitoring sensor network and storage medium | |
| CN112507479B (en) | Oil drilling machine health state assessment method based on manifold learning and softmax | |
| CN112668526A (en) | Bolt group loosening positioning monitoring method based on deep learning and piezoelectric active sensing | |
| CN116821804A (en) | Fracturing pump check valve fault on-line monitoring method based on multi-feature fusion | |
| CN113962308A (en) | Aviation equipment fault prediction method | |
| CN116738192A (en) | Digital twinning-based security data evaluation method and system | |
| CN114118219A (en) | Data-driven real-time abnormal detection method for health state of long-term power-on equipment | |
| CN118051822A (en) | Equipment running state abnormality detection method based on voiceprint recognition technology | |
| CN114330515A (en) | Bridge monitoring data abnormity diagnosis and repair method | |
| CN114818817A (en) | Weak fault recognition system and method for capacitive voltage transformer | |
| CN117786531A (en) | Bridge monitoring data anomaly diagnosis and repair system and method | |
| Dang et al. | seq2graph: Discovering dynamic non-linear dependencies from multivariate time series | |
| CN110850837A (en) | System life analysis and fault diagnosis method based on long-time and short-time memory neural network | |
| CN117667495B (en) | Association rule and deep learning integrated application system fault prediction method | |
| CN118050591A (en) | Distribution network 10KV overhead line fault early warning and positioning system based on AI | |
| CN117630797A (en) | Ammeter health state detection method, system and storage medium based on working current | |
| CN114638039B (en) | Structural health monitoring characteristic data interpretation method based on low-rank matrix recovery |
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 |