Transformer operation state monitoring system and method based on multi-dimensional data

Through multi-dimensional data monitoring system and artificial intelligence analysis, the problem that traditional monitoring methods cannot be monitored in real time and rely on manual operations is solved, and the automatic evaluation and accurate monitoring of the operating status of the transformer is realized to ensure the stability of the power system.

CN118914714AInactive Publication Date: 2024-11-08JIANGSU TIANJIN ELECTRIC POWER EQUIPMENT CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411052079.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-01
Publication Date
2024-11-08
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The traditional transformer monitoring method relies on regular maintenance and cannot achieve real-time monitoring. The excessive reliance on manual operations leads to high costs and errors, affecting the stable operation of the power system.

Method used

A transformer operating status monitoring system based on multi-dimensional data is adopted. By obtaining the dissolved gas concentration in the transformer oil, the top layer oil temperature, winding temperature and core grounding current value, artificial intelligence technology is used for feature encoding and analysis, and the transformer operating status is judged in real time.

Benefits of technology

It realizes automatic evaluation and real-time monitoring of the operating status of the transformer, improves the accuracy and reliability of monitoring, reduces manual intervention, promptly detects potential faults, and ensures the safety and stability of the power system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118914714A_ABST
    Figure CN118914714A_ABST
Patent Text Reader

Abstract

The invention provides a transformer operation state monitoring system and method based on multi-dimensional data, relates to the field of transformer intelligent monitoring, and adopts a data analysis technology based on artificial intelligence. And whether the operation state of the transformer is normal or not is judged by analyzing the concentration value of the dissolved gas in the transformer oil, the temperature value of the top oil of the transformer, the temperature value of the winding of the transformer and the grounding current value of the iron core of the transformer at multiple time points in real time. Therefore, automatic evaluation of the operation state of the transformer is realized, and the operation state of the transformer can be reflected more comprehensively through real-time monitoring of the multi-dimensional data of the transformer, so that the monitoring accuracy and reliability are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of intelligent transformer detection, and more specifically, to a transformer operating status monitoring system and method based on multi-dimensional data. Background Art

[0002] In today's era, with the continuous expansion of the scale of the power system and the significant increase in power load, the role of power transformers in the energy field has become increasingly critical and vital. To ensure that the power system can operate safely and stably, it is particularly important to monitor the operating status of power transformers. Traditional monitoring methods have always relied on regular inspections and maintenance, but this method faces many difficult problems. For example, the monitoring cycle is too long, resulting in the inability to detect potential fault hazards in a timely manner; the high dependence on manual labor not only consumes a lot of manpower costs, but also manual operations are inevitably prone to omissions and errors. These problems have brought certain challenges and risks to the stable operation of the power system in actual operation.

[0003] Therefore, an optimized transformer operation status monitoring solution based on multi-dimensional data is needed. Summary of the invention

[0004] In order to solve the above technical problems, the present application is proposed. The embodiments of the present application provide a transformer operation status monitoring system and method based on multi-dimensional data.

[0005] According to one aspect of the present application, a transformer operation status monitoring system based on multi-dimensional data is provided, which includes:

[0006] A transformer-related data acquisition module is used to obtain the concentration value of dissolved gas in transformer oil, the transformer top oil temperature value, the transformer winding temperature value and the transformer core grounding current value at multiple predetermined time points within a predetermined time period, wherein the dissolved gas in the transformer oil includes ethane, methane, hydrogen, ethylene, carbon monoxide and carbon dioxide;

[0007] A transformer-related data feature encoding module is used to feature encode the concentration values ​​of dissolved gas in transformer oil, the transformer top oil temperature value, the transformer winding temperature value, and the transformer core grounding current value at the plurality of predetermined time points to obtain a transformer operation status monitoring time series feature vector;

[0008] The transformer operation status monitoring result generation module is used to determine whether the transformer operation status is normal based on the transformer operation status monitoring time series feature vector.

[0009] According to another aspect of the present application, a transformer operating status monitoring method based on multidimensional data is provided, which includes:

[0010] Obtaining concentration values ​​of dissolved gases in transformer oil, transformer top oil temperature values, transformer winding temperature values, and transformer core grounding current values ​​at multiple predetermined time points within a predetermined time period, wherein the dissolved gases in the transformer oil include ethane, methane, hydrogen, ethylene, carbon monoxide, and carbon dioxide;

[0011] Characteristic encoding is performed on the concentration values ​​of dissolved gas in transformer oil, the top oil temperature value of transformer, the temperature value of transformer winding and the grounding current value of transformer core at the plurality of predetermined time points to obtain a time series characteristic vector for monitoring the transformer operation state;

[0012] Based on the transformer operation status monitoring time series feature vector, it is determined whether the transformer operation status is normal.

[0013] Compared with the prior art, the transformer operation status monitoring system and method based on multi-dimensional data provided by the present application adopts data analysis technology based on artificial intelligence, and performs real-time analysis on the concentration of dissolved gas in transformer oil, the top oil temperature of transformer, the temperature of transformer windings, and the grounding current value of transformer core at multiple time points to judge whether the transformer operation status is normal. In this way, not only the automatic evaluation of the transformer operation status is realized, but also the real-time monitoring of the transformer multi-dimensional data can more comprehensively reflect the operation status of the transformer, thereby improving the accuracy and reliability of monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0015] Figure 1 The present invention is a flowchart of a transformer operating status monitoring method based on multi-dimensional data according to an embodiment of the present application.

[0016] Figure 2 In the transformer operating status monitoring method based on multidimensional data according to an embodiment of the present application, feature encoding is performed on the concentration values ​​of dissolved gas in the transformer oil, the transformer top oil temperature value, the transformer winding temperature value and the transformer core grounding current value at the multiple predetermined time points to obtain a flowchart of the transformer operating status monitoring time series feature vector.

[0017] Figure 3In the transformer operating status monitoring method based on multidimensional data according to an embodiment of the present application, time series feature extraction is performed on the concentration values ​​of dissolved gas in the transformer oil at the multiple predetermined time points to obtain a flow chart of the time series feature vector of the dissolved gas concentration in the transformer oil.

[0018] Figure 4 In the transformer operating status monitoring method based on multidimensional data according to an embodiment of the present application, time series feature extraction is performed on the transformer top oil temperature value and the transformer winding temperature value at the plurality of predetermined time points to obtain a flow chart of the transformer key part temperature monitoring time series feature vector.

[0019] Figure 5 A flowchart of extracting the time series characteristics of the transformer core grounding current values ​​at the plurality of predetermined time points to obtain the transformer core grounding current time series characteristic vector in a transformer operating status monitoring method based on multi-dimensional data according to an embodiment of the present application.

[0020] Figure 6 This is a flowchart for judging whether the transformer operating state is normal based on the transformer operating state monitoring time series feature vector in the transformer operating state monitoring method based on multi-dimensional data according to an embodiment of the present application.

[0021] Figure 7 4 is a system block diagram of a transformer operation status monitoring system based on multi-dimensional data according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0023] In today's society, with the rapid development of the economy and the continuous advancement of technology, the continuous expansion of the scale of the power system and the significant increase in power load have become the norm. In this context, power transformers play an increasingly critical role in the energy field. Transformers are not only the core equipment for power transmission and distribution, but also the key link to ensure the safe and stable operation of the power system. Therefore, the monitoring of the operating status of power transformers is particularly important.

[0024] However, traditional monitoring methods, such as regular inspection and maintenance, although able to detect and prevent problems to a certain extent, have many limitations. First, the fixed nature of the monitoring cycle makes it impossible to achieve real-time monitoring of the transformer status, which may result in some sudden faults not being discovered and handled in a timely manner. Secondly, over-reliance on manual operation not only increases labor costs, but also makes it difficult to avoid omissions and errors caused by human factors. These defects have already posed certain challenges and risks to the stable operation of the power system in actual operation. Therefore, an optimized transformer operation status monitoring solution based on multidimensional data is expected.

[0025] In recent years, deep learning and neural networks have been widely used in computer vision, natural language processing, text signal processing and other fields. In addition, deep learning and neural networks have also shown a level close to or even beyond that of humans in image classification, object detection, semantic segmentation, text translation and other fields. The development of deep learning and neural networks provides new solutions and solutions for transformer operation status monitoring based on multi-dimensional data.

[0026] Figure 1 Flow chart of a transformer operating status monitoring method based on multi-dimensional data according to an embodiment of the present application. Figure 1 As shown, according to the transformer operation status monitoring method based on multidimensional data of an embodiment of the present application, the method includes: S110, obtaining the concentration value of dissolved gas in transformer oil, the transformer top oil temperature value, the transformer winding temperature value and the transformer core grounding current value at multiple predetermined time points within a predetermined time period, wherein the dissolved gas in the transformer oil includes ethane, methane, hydrogen, ethylene, carbon monoxide and carbon dioxide; S120, feature encoding the concentration value of dissolved gas in transformer oil, the transformer top oil temperature value, the transformer winding temperature value and the transformer core grounding current value at the multiple predetermined time points to obtain a transformer operation status monitoring time series feature vector; S130, judging whether the transformer operation status is normal based on the transformer operation status monitoring time series feature vector.

[0027] Specifically, in the technical solution of the present application, first, the concentration value of the dissolved gas in the transformer oil, the transformer top oil temperature value, the transformer winding temperature value and the transformer core grounding current value at multiple predetermined time points within a predetermined time period are obtained, and the dissolved gas in the transformer oil includes ethane, methane, hydrogen, ethylene, carbon monoxide and carbon dioxide. It should be understood that the transformer oil, as an insulating and cooling medium, will dissolve or produce various gases (such as ethane, methane, hydrogen, ethylene, carbon monoxide and carbon dioxide) during the operation of the transformer. The generation and concentration changes of these gases are often directly related to the insulation aging, overheating, partial discharge and other fault phenomena inside the transformer. The top oil temperature is an important indicator reflecting the thermal state of the transformer. The transformer will generate heat during operation. If the heat dissipation is poor or the load is too large, the top oil temperature will rise. Continuous high temperature will accelerate the aging of the insulation material and even cause failure. The winding is the main component in the transformer that bears electromagnetic force and current, and its temperature directly affects the operating efficiency and life of the transformer. The high winding temperature may be caused by excessive load, cooling system failure or insulation damage. The iron core is an important part of the transformer, and the magnitude of its grounding current reflects the insulation condition of the iron core. If the iron core is grounded at multiple points or the insulation is damaged, the grounding current will increase significantly. In general, there is a direct relationship between the monitoring parameters such as the concentration of dissolved gas in the transformer oil, the top oil temperature of the transformer, the temperature of the transformer winding, and the grounding current of the transformer core and the operating status and potential faults of the transformer. By monitoring the changes of these parameters in real time, the abnormal conditions of the transformer can be discovered in time, providing guarantee for the safe and stable operation of the power system.

[0028] In an embodiment of the present application, concentration values ​​of dissolved gases in transformer oil, transformer top oil temperature values, transformer winding temperature values, and transformer core grounding current values ​​at multiple predetermined time points within a predetermined time period are obtained, and the dissolved gases in the transformer oil include ethane, methane, hydrogen, ethylene, carbon monoxide, and carbon dioxide. One possible implementation method may be: using a gas chromatograph to obtain concentration data of different gases dissolved in the transformer oil; installing temperature sensors on the transformer top oil and windings to obtain corresponding temperature values ​​in real time; and installing a current transformer in the core grounding loop to monitor the grounding current.

[0029] Figure 2 In the transformer operation status monitoring method based on multi-dimensional data according to the embodiment of the present application, the concentration value of the dissolved gas in the transformer oil, the transformer top oil temperature value, the transformer winding temperature value and the transformer core grounding current value at the plurality of predetermined time points are feature-encoded to obtain a flow chart of the transformer operation status monitoring time series feature vector. Figure 2As shown, the concentration values ​​of dissolved gas in transformer oil, transformer top oil temperature, transformer winding temperature and transformer core grounding current at the plurality of predetermined time points are feature encoded to obtain a transformer operation status monitoring time series feature vector, including: S121, extracting time series features of the concentration values ​​of dissolved gas in transformer oil at the plurality of predetermined time points to obtain a time series feature vector of dissolved gas concentration in transformer oil; S122, extracting time series features of the transformer top oil temperature and transformer winding temperature at the plurality of predetermined time points to obtain a transformer key part temperature monitoring time series feature vector; S123, extracting time series features of the transformer core grounding current values ​​at the plurality of predetermined time points to obtain a transformer core grounding current time series feature vector; S124, fusing the time series feature vector of dissolved gas concentration in transformer oil, the time series feature vector of transformer key part temperature monitoring and the time series feature vector of transformer core grounding current to obtain the transformer operation status monitoring time series feature vector.

[0030] Figure 3 The present invention is a flowchart of extracting time series characteristics of the concentration values ​​of the dissolved gas in the transformer oil at the plurality of predetermined time points in the transformer operating state monitoring method based on multidimensional data according to an embodiment of the present application to obtain a time series characteristic vector of the concentration of the dissolved gas in the transformer oil. Figure 3 As shown, the time series feature extraction of the concentration values ​​of dissolved gas in transformer oil at the multiple predetermined time points to obtain the time series feature vector of dissolved gas concentration in transformer oil includes: S1211, the concentration values ​​of dissolved gas in transformer oil at the multiple predetermined time points are extracted according to the time dimension to obtain a time series input matrix of dissolved gas concentration in transformer oil; S1212, the time series input matrix of dissolved gas concentration in transformer oil is passed through a dissolved gas time series encoder to obtain a time series feature matrix of dissolved gas concentration in transformer oil; S1213, the time series feature matrix of dissolved gas concentration in transformer oil is expanded into the time series feature vector of dissolved gas concentration in transformer oil.

[0031] Next, the concentration values ​​of the dissolved gas in the transformer oil at the multiple predetermined time points are arranged according to the time dimension to obtain a time series input matrix of the concentration of dissolved gas in the transformer oil. It should be understood that, considering that the concentration of dissolved gas in the transformer oil changes over time, this change is often closely related to the operating status and potential faults of the transformer. In order to use time series analysis technology to capture this dynamic change, thereby revealing the changing trend of the transformer state, in the technical solution of the present application, the concentration values ​​of the dissolved gas in the transformer oil at the multiple predetermined time points are arranged according to the time dimension. In the time series input matrix of the concentration of dissolved gas in the transformer oil formed by the arrangement, each row is the data of the concentration value of dissolved gas in the transformer oil at a specific time point, and each column is the data of a specific concentration of a dissolved gas parameter indicator.

[0032] Considering that the time series data of the concentration of dissolved gas in transformer oil is an important indicator reflecting the operating state of the transformer, the time series data formed by the original arrangement contains some redundant information and noise, and it is difficult to directly capture the local dependence of the dissolved gas concentration over time, so it is difficult to characterize the operating state of the transformer. Therefore, it is necessary to pass the dissolved gas concentration time series input matrix in the transformer oil through the dissolved gas time series encoder to obtain the dissolved gas concentration time series feature matrix in the transformer oil. The dissolved gas time series encoder in this application is a convolutional neural network model using non-overlapping convolution kernels. Ordinary technicians in this field should know that the convolutional neural network model has a strong feature extraction capability and can automatically learn and extract useful feature representations from the input data. The convolutional neural network model performs sliding operations on the data through the convolution kernel to realize feature extraction of the local area. The technical solution of this application uses a non-overlapping convolution kernel, and its moving step length on the input data is equal to its width, that is, each convolution operation will not overlap. This means that when the time series input matrix of the dissolved gas concentration in the transformer oil is operated, the amount of calculation is greatly reduced and the processing speed of the model is improved, because each position is only convolved once.

[0033] In an embodiment of the present application, a possible implementation method of passing the time series input matrix of dissolved gas concentration in transformer oil through a dissolved gas time series encoder to obtain a time series feature matrix of dissolved gas concentration in transformer oil can be: each layer of the dissolved gas time series encoder performs the following on the input data during the forward transfer process of the layer: convolving the input data based on the non-overlapping convolution kernel to generate a convolution feature map, wherein the width of the non-overlapping convolution kernel is equal to the moving step of the non-overlapping convolution kernel on the input data; performing pooling processing on the convolution feature map along the channel dimension to obtain a down-sampled feature map; performing activation processing on the down-sampled feature map based on a nonlinear activation function; wherein the input of the first layer of the dissolved gas time series encoder is the time series input matrix of dissolved gas concentration in transformer oil, and the output of the last layer is the time series feature matrix of dissolved gas concentration in transformer oil.

[0034] Next, the time series characteristic matrix of the concentration of dissolved gas in the transformer oil is expanded into a time series characteristic vector of the concentration of dissolved gas in the transformer oil. It should be understood that by expanding the characteristic matrix into a vector, the dimension reduction of the characteristic data is achieved, the complexity of the data is reduced, and the computing resources required for the model to process the data are reduced. It is worth noting that although the characteristic matrix is ​​expanded into a vector, the continuity of the time series information is still retained during the expansion process, which means that the characteristic vector can still reflect the dynamic changes of the concentration of dissolved gas in the transformer oil.

[0035] Figure 4 In the transformer operation status monitoring method based on multi-dimensional data according to the embodiment of the present application, the time series feature extraction is performed on the transformer top oil temperature value and the transformer winding temperature value at the plurality of predetermined time points to obtain the flow chart of the transformer key part temperature monitoring time series feature vector. Figure 4 As shown, the time series feature extraction of the transformer top oil temperature value and the transformer winding temperature value at the plurality of predetermined time points to obtain the transformer key part temperature monitoring time series feature vector includes: S1221, arranging the transformer top oil temperature value and the transformer winding temperature value at the plurality of predetermined time points into a transformer key part temperature monitoring time series input matrix according to the time dimension and the temperature sample dimension; S1222, passing the transformer key part temperature monitoring time series input matrix through a transformer temperature timing encoder to obtain a transformer key part temperature monitoring time series feature matrix; S1223, expanding the transformer key part temperature monitoring time series feature matrix into the transformer key part temperature monitoring time series feature vector.

[0036] At the same time, the transformer top oil temperature values ​​and transformer winding temperature values ​​at the multiple predetermined time points are arranged according to the time dimension and the temperature sample dimension as a transformer key part temperature monitoring timing input matrix. Accordingly, considering that the transformer top oil temperature and the winding temperature are two important parameters reflecting the transformer operating state, in order to comprehensively consider the changes of the two, so as to more comprehensively reflect the transformer operating state, in the technical solution of the present application, the transformer top oil temperature values ​​and transformer winding temperature values ​​at the multiple predetermined time points are arranged according to the time dimension and the temperature sample dimension. The arrangement is formed to obtain the transformer key part temperature monitoring timing input matrix. This matrix data expression method can centrally represent discrete data and improve information density, thereby facilitating subsequent temperature feature extraction.

[0037] The temperature data of the two key parts of the transformer, the top oil temperature data of the transformer and the transformer winding temperature data, change over time. This time series data contains rich information, such as the temperature change trend, fluctuation range, etc. In order to capture the temperature time series characteristics in these data, so as to indirectly understand the internal state of the transformer, the transformer key part temperature monitoring time series input matrix is ​​passed through the transformer temperature time series encoder to obtain the transformer key part temperature monitoring time series feature matrix. The essence of the transformer temperature time series encoder in this application is a convolutional neural network model using a spatial attention mechanism. The spatial attention mechanism allows the model to pay more attention to certain important parts in the image or time series when processing the input data. In transformer temperature monitoring, this means that the model can automatically identify and pay attention to the temperature changes of key parts that have a greater impact on the operating state of the transformer. By introducing the spatial attention mechanism, the CNN model can more accurately capture the details and trends of temperature changes when processing the transformer temperature time series data, thereby improving the accuracy of judging the operating state of the transformer.

[0038] In order to convert the high-dimensional temperature time series characteristic matrix data into lower-dimensional data that is easier to process, in the technical solution of the present application, the transformer key part temperature monitoring time series characteristic matrix is ​​expanded into a transformer key part temperature monitoring time series characteristic vector. In particular, in a specific embodiment of the present application, the characteristic matrix can be expanded row by row, and the elements of each row can be arranged in sequence to form a characteristic vector.

[0039] Figure 5 A flowchart of extracting the time series characteristics of the transformer core grounding current values ​​at the plurality of predetermined time points to obtain the transformer core grounding current time series characteristic vector in a transformer operation status monitoring method based on multi-dimensional data according to an embodiment of the present application. Figure 5As shown, the timing feature extraction of the transformer core grounding current values ​​at the multiple predetermined time points to obtain the transformer core grounding current timing feature vector includes: S1231, arranging the transformer core grounding current values ​​at the multiple predetermined time points into a transformer core grounding current timing input vector according to the time dimension; S1232, passing the transformer core grounding current timing input vector through a transformer grounding current timing encoder to obtain the transformer core grounding current timing feature vector.

[0040] Considering that the core grounding current of the transformer will change dynamically over time during operation, this change may be affected by various factors, such as load changes, internal faults, etc. In order to intuitively reflect these dynamic changes, it is necessary to arrange the transformer core grounding current values ​​at the multiple predetermined time points according to the time dimension as a transformer core grounding current time series input vector. By arranging according to the time dimension, the time series information in the current data can be retained.

[0041] By analyzing the time series changes of the transformer core grounding current, abnormal fluctuations or mutations in the current can be discovered in time. These anomalies are often related to internal faults or abnormal operation of the transformer. Therefore, the transformer core grounding current time series input vector is passed through the transformer grounding current time series encoder to obtain the transformer core grounding current time series feature vector. In this application, the transformer grounding current time series encoder is a convolutional neural network model using a one-dimensional convolution kernel. It is suitable for processing one-dimensional time series data and can effectively capture the time series characteristics and local correlations in the current data. By performing a sliding window operation on the time series data through the one-dimensional convolution kernel in the convolution layer, the key features in the current data can be automatically extracted, and because the convolution operation has a certain invariance to the local transformation of the input data (such as translation, rotation, etc.), the convolutional neural network using a one-dimensional convolution kernel has good robustness when processing time series data with noise or slight disturbances.

[0042] Considering that the concentration of dissolved gas in transformer oil, the temperature of key parts and the core grounding current respectively reflect the different physical and chemical states inside the transformer, there is a certain complementarity between these parameters, that is, they each provide different side information of the transformer operating state. In order to have a more comprehensive understanding of the overall operating state of the transformer, the time series feature vector of the concentration of dissolved gas in transformer oil, the time series feature vector of the temperature monitoring of key parts of the transformer and the time series feature vector of the transformer core grounding current are integrated in the technical solution of the present application to obtain the time series feature vector of the transformer operating state monitoring. It should be well understood that single data is affected by noise, interference or measurement error, which will lead to inaccurate monitoring of the transformer operating state. In the present application, by integrating feature vectors from multi-source data, the information between them can be used to verify and correct each other, thereby improving the accuracy and reliability of transformer operating state monitoring.

[0043] Figure 6 This is a flowchart of judging whether the transformer operating state is normal based on the transformer operating state monitoring time series feature vector in the transformer operating state monitoring method based on multi-dimensional data according to an embodiment of the present application. Figure 6 As shown, judging whether the transformer operating state is normal based on the transformer operating state monitoring time series feature vector includes: S131, performing a compatibility evaluation on the transformer operating state monitoring time series feature vector based on an instance disturbance differential adjustment variable to obtain an optimized transformer operating state monitoring time series feature vector; S132, passing the optimized transformer operating state monitoring time series feature vector through a transformer operating state classifier to obtain a classification result, and the classification result is used to indicate whether the transformer operating state is normal.

[0044] In particular, in the actual transformer-related data collection process, due to sensor accuracy, environmental factors, equipment aging and other issues, data collection errors will occur, and because the operating environment of the transformer (such as temperature, humidity, etc.) will change, these changes will affect the operating state of the transformer to a certain extent, which will cause the transformer operating state monitoring time series feature vector obtained by encoding the transformer operating data to have a special instance disturbance of the eigenvalue. In the technical solution of the present application, considering that the neural network model performs feature mining on the input transformer operating state data based on the network parameter matrix, if the inherent implicit association between the network output feature and the network parameter matrix can be further considered, the prior information of the neural network model can be further used to modulate the characteristic manifold of the transformer operating state monitoring time series feature vector to reduce the possibility of the occurrence of special instance disturbances of the eigenvalue. That is, the transformer operating state monitoring time series feature vector is evaluated for compatibility based on the differential adjustment variable of the instance disturbance to obtain the optimized transformer operating state monitoring time series feature vector.

[0045] Specifically, in an embodiment of the present application, a compatibility evaluation is performed on the transformer operation state monitoring time series feature vector based on an instance disturbance differential adjustment variable to obtain an optimized transformer operation state monitoring time series feature vector, including: calculating the mean and variance of the transformer operation state monitoring time series feature vector; calculating the difference between the eigenvalue of the transformer operation state monitoring time series feature vector at a predetermined position and the mean of the transformer operation state monitoring time series feature vector to obtain a first intermediate calculated value; multiplying the network parameter matrix by the transformer operation state monitoring time series feature vector and then subtracting the network parameter matrix from the transformer operation state monitoring time series feature vector by position to obtain a first intermediate vector, and calculating the Frobenius norm of the first intermediate vector to obtain a first norm value; The norm value is added to the variance of the transformer operation status monitoring time series characteristic vector and then a square root operation is performed to obtain a second intermediate calculation value; the first intermediate calculation value is divided by the second intermediate calculation value to obtain a third intermediate calculation value; the squares of each position of the transformer operation status monitoring time series characteristic vector are calculated and then accumulated to obtain a first sum value; the square root of the first sum value is divided by the scale of the network parameter matrix and then the natural exponential function value is calculated to obtain a fourth intermediate calculation value, wherein the scale of the network parameter matrix is ​​equal to the product between the width value of the network parameter matrix and the height value thereof; the fourth intermediate calculation value is multiplied by the weighted hyperparameter and then added to the third intermediate calculation value to obtain the characteristic value of the predetermined position of the optimized transformer operation status monitoring time series characteristic vector.

[0046] In an embodiment of the present application, another expression of performing a compatibility evaluation on the transformer operation state monitoring time series feature vector based on the instance disturbance differential adjustment variable to obtain an optimized transformer operation state monitoring time series feature vector can be: processing the transformer operation state monitoring time series feature vector according to the following formula to obtain the optimized transformer operation state monitoring time series feature vector; wherein the formula is:

[0047]

[0048] Where V is the transformer operating status monitoring time series feature vector, M is the network parameter matrix, β represents the weighted hyperparameter, v i is the eigenvalue of the ith position of the transformer operation status monitoring time series feature vector, N is the scale of the network parameter matrix, the scale of the network parameter matrix is ​​equal to the product of the width value and the height value of the network parameter matrix, L is the length of the transformer operation status monitoring time series feature vector, ||·|| F represents the F norm of the feature vector, μ represents the mean value of the transformer operation status monitoring time series feature vector, σ 2represents the variance of the transformer operating status monitoring time series feature vector, exp(·) represents the exponential function with the natural constant e as the base, v' i represents the eigenvalue of the i-th position of the optimized transformer operation status monitoring time series eigenvector, Represents matrix multiplication.

[0049] It should be understood that in the technical solution of the present application, considering that the neural network model performs feature mining on the input data based on the network parameter matrix, if the inherent implicit association between the network output feature and the network parameter matrix can be further considered, the prior information of the neural network model can be further used to modulate the characteristic manifold of the transformer operation state monitoring time series feature vector to reduce the possibility of the occurrence of special instance disturbances of its characteristic values. Based on this, in a preferred embodiment, the transformer operation state monitoring time series feature vector is evaluated for compatibility based on the instance disturbance differential adjustment variable, which first maps the transformer operation state monitoring time series feature vector to the high-dimensional feature space where the network parameter matrix is ​​located, and uses the compatibility perception factor between the mapping result and the original input as the inherent implicit association descriptor between the network output feature and the network parameter matrix. Then, the statistical characteristics of the feature group of the transformer operation state monitoring time series feature vector are used as the inherent modal information of the transformer operation state monitoring time series feature vector and the inherent implicit association descriptor is used as the instance disturbance differential adjustment variable to perform data dynamic standardization on the attribute values ​​of each position of the transformer operation state monitoring time series feature vector. At the same time, the class-standardized modulus data of the transformer operating status monitoring time series feature vector is used as a bias item to constrain the acceptable range of the population distribution of the transformer operating status monitoring time series feature vector. In this way, the attribute influence of the interference distribution of abnormal samples can be suppressed as much as possible, thereby improving the expression accuracy of the transformer operating status monitoring time series feature vector.

[0050] Finally, the optimized transformer operation state monitoring time series feature vector is passed through the transformer operation state classifier to obtain a classification result, and the classification result is used to indicate whether the transformer operation state is normal. A classifier is a machine learning model, and generally transformer data of known state (i.e., a labeled data set) is used to train the classifier. These data sets should contain transformer-related data in normal state and multiple fault states, as well as corresponding labels (such as "normal", "overheating", "insulation fault", etc.). Through training, the classifier can learn the distribution law and feature differences of feature vectors under different states. During the training process, the parameters of the classifier need to be tuned to improve the accuracy and generalization ability of the classification, including the adjustment of hyperparameters such as the learning rate, the number of iterations, and the regularization term. The transformer operation state monitoring time series feature vector to be monitored is input into the trained classifier, and the classifier performs classification judgment based on the input feature vector and the knowledge learned during the training process. The classification result is directly used to indicate whether the current operation state of the transformer is normal. The operation and maintenance personnel can timely understand the health status of the transformer according to the classification result, and take corresponding measures for maintenance or overhaul.

[0051] In summary, the transformer operation status monitoring method based on multi-dimensional data according to the embodiment of the present application is explained, which adopts data analysis technology based on artificial intelligence, and judges whether the transformer operation status is normal by real-time analysis of the dissolved gas concentration value in the transformer oil, the transformer top oil temperature value, the transformer winding temperature value and the transformer core grounding current value at multiple time points. In this way, not only the automatic evaluation of the transformer operation status is realized, but also the real-time monitoring of the transformer multi-dimensional data can more comprehensively reflect the operation status of the transformer, thereby improving the accuracy and reliability of monitoring.

[0052] Figure 7 : is a system block diagram of a transformer operating status monitoring system based on multidimensional data according to an embodiment of the present application. Figure 7As shown, according to the transformer operation status monitoring system 100 based on multidimensional data of the embodiment of the present application, it includes: a transformer-related data acquisition module 110, which is used to obtain the concentration value of dissolved gas in transformer oil, the transformer top oil temperature value, the transformer winding temperature value and the transformer core grounding current value at multiple predetermined time points within a predetermined time period, wherein the dissolved gas in the transformer oil includes ethane, methane, hydrogen, ethylene, carbon monoxide and carbon dioxide; a transformer-related data feature encoding module 120, which is used to feature encode the concentration value of dissolved gas in transformer oil, the transformer top oil temperature value, the transformer winding temperature value and the transformer core grounding current value at the multiple predetermined time points to obtain a transformer operation status monitoring time series feature vector; a transformer operation status monitoring result generation module 130, which is used to determine whether the transformer operation status is normal based on the transformer operation status monitoring time series feature vector.

[0053] Here, those skilled in the art can understand that the specific functions and operations of the various units and modules in the transformer operation status monitoring system 100 based on multidimensional data have been described in the above reference. Figures 1 to 6 The description of transformer operation status monitoring based on multi-dimensional data has been introduced in detail, and therefore, its repeated description will be omitted.

[0054] In summary, the transformer operation status monitoring system 100 based on multi-dimensional data according to the embodiment of the present application is explained, which adopts data analysis technology based on artificial intelligence, and judges whether the transformer operation status is normal by performing real-time analysis on the dissolved gas concentration value in the transformer oil, the transformer top oil temperature value, the transformer winding temperature value and the transformer core grounding current value at multiple time points. In this way, not only the automatic evaluation of the transformer operation status is realized, but also the real-time monitoring of the transformer multi-dimensional data can more comprehensively reflect the operation status of the transformer, thereby improving the accuracy and reliability of the monitoring.

[0055] As described above, the transformer operation status monitoring system 100 based on multidimensional data according to the embodiment of the present application can be implemented in various wireless terminals, such as a server for transformer operation status monitoring based on multidimensional data. In one example, the transformer operation status monitoring system 100 based on multidimensional data according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the transformer operation status monitoring system 100 based on multidimensional data can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the transformer operation status monitoring system 100 based on multidimensional data can also be one of the many hardware modules of the wireless terminal.

[0056] Alternatively, in another example, the transformer operation status monitoring system 100 based on multi-dimensional data and the wireless terminal may also be separate devices, and the transformer operation status monitoring system 100 based on multi-dimensional data may be connected to the wireless terminal via a wired and / or wireless network, and transmit interactive information in accordance with an agreed data format.

Claims

1. A transformer operating status monitoring method based on multidimensional data, characterized in that: include: Obtaining concentration values ​​of dissolved gases in transformer oil, transformer top oil temperature values, transformer winding temperature values, and transformer core grounding current values ​​at multiple predetermined time points within a predetermined time period, wherein the dissolved gases in the transformer oil include ethane, methane, hydrogen, ethylene, carbon monoxide, and carbon dioxide; Characteristic encoding is performed on the concentration values ​​of dissolved gas in transformer oil, the top oil temperature value of transformer, the temperature value of transformer winding and the grounding current value of transformer core at the plurality of predetermined time points to obtain a time series characteristic vector for monitoring the transformer operation state; Based on the transformer operation status monitoring time series feature vector, it is determined whether the transformer operation status is normal.

2. The transformer operating status monitoring method based on multidimensional data according to claim 1 is characterized in that: The concentration values ​​of the dissolved gas in the transformer oil, the transformer top oil temperature value, the transformer winding temperature value, and the transformer core grounding current value at the plurality of predetermined time points are feature-encoded to obtain a transformer operation status monitoring time series feature vector, including: Performing time series feature extraction on the concentration values ​​of dissolved gas in the transformer oil at the plurality of predetermined time points to obtain a time series feature vector of the concentration of dissolved gas in the transformer oil; Extracting time series features of the transformer top oil temperature values ​​and transformer winding temperature values ​​at the plurality of predetermined time points to obtain a time series feature vector for monitoring the temperature of a key part of the transformer; Extracting time series characteristics of the transformer core grounding current values ​​at the plurality of predetermined time points to obtain a transformer core grounding current time series characteristic vector; The transformer operation status monitoring time series feature vector is obtained by integrating the time series feature vector of the concentration of dissolved gas in the transformer oil, the time series feature vector of the temperature monitoring of the key parts of the transformer and the time series feature vector of the transformer core grounding current.

3. The transformer operating status monitoring method based on multidimensional data according to claim 2 is characterized in that: Performing time series feature extraction on the concentration values ​​of dissolved gas in the transformer oil at the plurality of predetermined time points to obtain a time series feature vector of the concentration of dissolved gas in the transformer oil, comprising: The concentration values ​​of the dissolved gas in the transformer oil at the plurality of predetermined time points are input according to the time dimension to obtain a time series matrix of the concentration of the dissolved gas in the transformer oil; The dissolved gas concentration time series input matrix in the transformer oil is passed through a dissolved gas time series encoder to obtain a dissolved gas concentration time series characteristic matrix in the transformer oil; The time series characteristic matrix of the concentration of dissolved gas in the transformer oil is expanded into a time series characteristic vector of the concentration of dissolved gas in the transformer oil.

4. The transformer operating status monitoring method based on multidimensional data according to claim 3 is characterized in that: The time series feature extraction is performed on the transformer top oil temperature values ​​and transformer winding temperature values ​​at the plurality of predetermined time points to obtain a transformer key part temperature monitoring time series feature vector, including: Arrange the transformer top oil temperature values ​​and transformer winding temperature values ​​at the plurality of predetermined time points according to the time dimension and the temperature sample dimension into a transformer key part temperature monitoring timing input matrix; The transformer key part temperature monitoring timing input matrix is ​​passed through the transformer temperature timing encoder to obtain the transformer key part temperature monitoring timing characteristic matrix; The transformer key part temperature monitoring time series characteristic matrix is ​​expanded into the transformer key part temperature monitoring time series characteristic vector.

5. The transformer operating status monitoring method based on multidimensional data according to claim 4 is characterized in that: Extracting the time series characteristics of the transformer core grounding current values ​​at the plurality of predetermined time points to obtain a transformer core grounding current time series characteristic vector includes: Arranging the transformer core grounding current values ​​at the plurality of predetermined time points into a transformer core grounding current time series input vector according to the time dimension; The transformer core grounding current timing input vector is passed through a transformer grounding current timing encoder to obtain the transformer core grounding current timing characteristic vector.

6. The transformer operating status monitoring method based on multidimensional data according to claim 5 is characterized in that: The dissolved gas timing encoder is a convolutional neural network model that uses non-overlapping convolution kernels in adjacent layers, the transformer temperature timing encoder is a convolutional neural network model that uses a spatial attention mechanism, and the transformer grounding current timing encoder is a convolutional neural network model that uses a one-dimensional convolution kernel.

7. The transformer operating status monitoring method based on multidimensional data according to claim 6 is characterized in that: Based on the transformer operation status monitoring time series feature vector, judging whether the transformer operation status is normal includes: Performing a compatibility evaluation on the transformer operation state monitoring time series feature vector based on an instance disturbance differential adjustment variable to obtain an optimized transformer operation state monitoring time series feature vector; The optimized transformer operation status monitoring time series feature vector is passed through a transformer operation status classifier to obtain a classification result, and the classification result is used to indicate whether the transformer operation status is normal.

8. The transformer operating status monitoring method based on multidimensional data according to claim 7 is characterized in that: The transformer operation state monitoring time series feature vector is evaluated for compatibility based on an instance disturbance differential adjustment variable to obtain an optimized transformer operation state monitoring time series feature vector, including: Calculating the mean and variance of the transformer operation status monitoring time series feature vector; Calculate the difference between the characteristic value of the predetermined position of the transformer operation state monitoring time series characteristic vector and the mean value of the transformer operation state monitoring time series characteristic vector to obtain a first intermediate calculation value; Multiplying the network parameter matrix by the transformer operation state monitoring time series feature vector and then subtracting the matrix from the transformer operation state monitoring time series feature vector by position to obtain a first intermediate vector, and calculating the Frobenius norm of the first intermediate vector to obtain a first norm value; Adding the first norm value to the variance of the transformer operation status monitoring time series characteristic vector and then performing a square root operation to obtain a second intermediate calculation value; dividing the first intermediate calculated value by the second intermediate calculated value to obtain a third intermediate calculated value; Calculating the square of each position of the transformer operation status monitoring time series characteristic vector and then accumulating and summing them to obtain a first sum value; Calculate a natural exponential function value after taking the square root of the first sum value and dividing it by the scale of the network parameter matrix to obtain a fourth intermediate calculation value, wherein the scale of the network parameter matrix is ​​equal to the product of the width value of the network parameter matrix and the height value thereof; The fourth intermediate calculated value is multiplied by the weighted hyperparameter and then added to the third intermediate calculated value to obtain the characteristic value of the predetermined position of the optimized transformer operation status monitoring time series characteristic vector.

9. A transformer operation status monitoring system based on multidimensional data, characterized in that: include: A transformer-related data acquisition module is used to obtain the concentration value of dissolved gas in transformer oil, the transformer top oil temperature value, the transformer winding temperature value and the transformer core grounding current value at multiple predetermined time points within a predetermined time period, wherein the dissolved gas in the transformer oil includes ethane, methane, hydrogen, ethylene, carbon monoxide and carbon dioxide; A transformer-related data feature encoding module is used to feature encode the concentration values ​​of dissolved gas in transformer oil, the transformer top oil temperature value, the transformer winding temperature value, and the transformer core grounding current value at the plurality of predetermined time points to obtain a transformer operation status monitoring time series feature vector; The transformer operation status monitoring result generation module is used to determine whether the transformer operation status is normal based on the transformer operation status monitoring time series feature vector.

10. The transformer operation status monitoring system based on multi-dimensional data according to claim 9, characterized in that: The transformer-related data feature encoding module includes: A time series feature extraction unit for the concentration of dissolved gas in transformer oil, used for extracting the time series features of the concentration values ​​of dissolved gas in transformer oil at the plurality of predetermined time points to obtain a time series feature vector of the concentration of dissolved gas in transformer oil; A transformer key part temperature monitoring data time series feature extraction unit is used to extract the time series features of the transformer top oil temperature value and the transformer winding temperature value at the plurality of predetermined time points to obtain a transformer key part temperature monitoring time series feature vector; A transformer core grounding current time series feature extraction unit, configured to extract time series features of the transformer core grounding current values ​​at the plurality of predetermined time points to obtain a transformer core grounding current time series feature vector; The feature fusion unit is used to fuse the time series feature vector of the dissolved gas concentration in the transformer oil, the time series feature vector of the temperature monitoring of the key parts of the transformer and the time series feature vector of the transformer core grounding current to obtain the time series feature vector of the transformer operation status monitoring.