Transformer fault diagnosis method and system based on statistical characteristics of multiple monitoring variables

Abstract

The application discloses a transformer fault diagnosis method and system based on statistical characteristics of multiple monitoring variables, collects power transformer online monitoring data in a cycle, and performs data preprocessing on the power transformer online monitoring data; the power transformer online monitoring data after the preprocessing is subjected to distribution statistical model loading, and statistical law analysis is performed based on the loaded statistical distribution model to determine a distribution statistical model and model parameters; more cycles of the power transformer online monitoring data after the preprocessing are acquired, input into the determined distribution statistical model, and a plurality of groups of model parameters are obtained; the model parameters obtained according to the power transformer online monitoring data in different cycles are compared, if the errors between the model parameters are within a threshold, it is determined that the power transformer is faulty, otherwise, it is determined that the power transformer is faulty. The application also has a wide application prospect in power transformer operation maintenance, inspection, live detection and online monitoring.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of online monitoring of electrical engineering power transformers, and particularly relates to a transformer fault diagnosis method and system based on statistical characteristics of multiple monitoring parameters. BACKGROUND

[0002] At present, online monitoring of the insulation state of power transformers is the main method of equipment operation and maintenance management in the power industry, and has good real-time performance and accuracy. However, due to the complex field operation conditions of transformer equipment, the performance of state monitoring sensors, network communication quality and other factors, the transformer online monitoring data has the common problems of large fluctuation and many abnormal data, resulting in low quality of online monitoring data and low data reliability, and it is impossible to directly use the online monitoring data to analyze and judge the state of the transformer. At present, the analysis of online monitoring data of power transformers is limited to data analysis of a single monitoring parameter itself, such as oil dissolved gas analysis and partial discharge. These technologies cannot achieve horizontal mining of all online monitoring data of power transformers, and there is no correlation analysis between parameters combined with data distribution characteristics. The acquisition of online monitoring data of power transformers involves a huge amount of data, and it is necessary to parse unstructured data of different online monitoring into structured data, which has great difficulty in data volume itself, data acquisition and data parsing.

[0003] At present, the data of various online monitoring devices and systems cannot be obtained from public channels. Each online monitoring device manufacturer can only master the data of its own products, and can only speculate the running state of the equipment by accumulating data on site, which has strong one-sidedness, and the data obtained by this method for state evaluation of the monitored equipment is very different from the real data and is very unstable.

[0004] The existing online monitoring system often needs to rely on the setting of thresholds by human beings to judge the state of the monitored equipment. SUMMARY

[0005] The purpose of the present application is to provide a transformer fault diagnosis method and system based on statistical characteristics of multiple monitoring parameters to solve the defects of the prior art. The present application no longer determines the threshold of equipment monitoring data subjectively, and establishes correlation analysis rules through multi-parameter statistics of online monitoring data to realize state analysis and fault diagnosis of the monitored transformer. The present application also has a broad application prospect in the operation and maintenance, inspection, live detection and online monitoring of power transformers.

[0006] To achieve the above purpose, the present application adopts the following technical solutions:

[0007] The transformer fault diagnosis method based on statistical characteristics of multiple monitoring parameters comprises the following steps:

[0008] S100: Collecting power transformer online monitoring data of one cycle, and performing data preprocessing on the power transformer online monitoring data;

[0009] S200: Loading a distribution statistical model on the preprocessed power transformer online monitoring data, and performing statistical rule analysis based on the loaded statistical distribution model to determine the distribution statistical model and model parameters;

[0010] S300: Obtaining more cycles of preprocessed power transformer online monitoring data, inputting the distribution statistical model determined in S200, and obtaining a plurality of groups of model parameters;

[0011] S400: Comparing the model parameters obtained according to the power transformer online monitoring data of different cycles, if the error between the model parameters is within a threshold, determining that the power transformer is fault-free, otherwise determining that the power transformer has a fault.

[0012] Further, the power transformer online monitoring data includes oil dissolved gas data, partial discharge data, core grounding current data, oil temperature and oil pressure.

[0013] Further, the data preprocessing includes data cleaning, data integration, data transformation and data reduction.

[0014] Further, the data integration includes:

[0015] Solve the problem that the time index of the power transformer online monitoring data is not monotonically increasing;

[0016] Solve the problem that the time index of the power transformer online monitoring data contains repeated timestamps, by fixing the first value;

[0017] Solve the problem that the power transformer online monitoring data lacks the time index attribute freq;

[0018] Solve the problem that the time series of the power transformer online monitoring data includes a classification label column, by converting the classification label into a binary indicator to repair.

[0019] Further, the statistical distribution model includes a single-parameter statistical model and a multi-parameter statistical model.

[0020] Further, the single-parameter statistical model includes normal distribution, Weibull distribution and t-parameter distribution.

[0021] Further, the multi-parameter statistical model includes multiple linear regression detection, Logistic regression detection, Cox proportional risk regression detection and time series detection.

[0022] Further, the statistical rule analysis is based on Python language and utilizes Pandas package.

[0023] Further, the model parameter is a parameter capable of determining the distribution statistical model.

[0024] The transformer fault diagnosis system based on statistical characteristics of multiple monitoring variables comprises:

[0025] A data acquisition and preprocessing module is used for acquiring one cycle of power transformer online monitoring data and performing data preprocessing on the power transformer online monitoring data.

[0026] A model determination module is used for loading a distribution statistical model on the preprocessed power transformer online monitoring data, and performing statistical rule analysis based on the loaded statistical distribution model to determine the distribution statistical model and the model parameter.

[0027] A model parameter acquisition module is used for acquiring more cycles of preprocessed power transformer online monitoring data, inputting the determined distribution statistical model, and obtaining a plurality of groups of model parameters.

[0028] A judgment module is used for comparing the model parameters obtained according to different cycles of power transformer online monitoring data, and if the error between the model parameters is within a threshold, it is judged that the power transformer is fault-free, otherwise it is judged that the power transformer has a fault.

[0029] Compared with the prior art, the present application has the following beneficial technical effects:

[0030] The present application combines the statistical characteristics of power transformer online monitoring data, utilizes the joint statistical analysis algorithm between multiple cycles of power transformer online monitoring data, dynamically fuses and analyzes the running state of the monitored power transformer, and can obtain very high accuracy of power transformer state analysis and fault diagnosis; the present application no longer determines the equipment monitoring data threshold subjectively, establishes the correlation analysis rule through the multi-parameter statistics of online monitoring data, so as to realize the state analysis and fault diagnosis of the monitored transformer, and has a wide application prospect in power transformer operation and maintenance, inspection, live detection and online monitoring. BRIEF DESCRIPTION OF DRAWINGS

[0031] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the present application.

[0032] Figure 1 The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the present application.

[0033] Figure 2The specific flow chart of the single parameter data analysis method for the on-line monitoring of the power transformer according to the present application is shown in the figure;

[0034] Figure 3 The specific flow chart of the multi-parameter data analysis method for the on-line monitoring of the power transformer according to the present application is shown in the figure;

[0035] Figure 4 The statistical characteristic example graph of the single state parameter according to the present application is shown in the figure, wherein (a) is the H2 content statistics in the oil before the partial discharge, and (b) is the H2 content statistics in the oil after the partial discharge;

[0036] Figure 5 The fitting scatter plot of the multi-state monitoring parameter according to the present application is shown in the figure;

[0037] Figure 6 The flow chart of the transformer fault diagnosis method based on the statistical characteristics of the multi-monitoring parameter according to the present application is shown in the figure. DETAILED DESCRIPTION

[0038] In order for those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings of the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should fall within the scope of the present application.

[0039] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to the clearly listed steps or units, but can include other steps or units not clearly listed or inherent to the process, method, product or device.

[0040] Embodiment one

[0041] The present application provides a transformer fault diagnosis method based on the statistical characteristics of multi-monitoring parameters, which combines the statistical characteristics of the on-line monitoring data of the power transformer, and uses the joint statistical analysis algorithm between the multi-cycle on-line monitoring data of the power transformer to dynamically fuse and diagnose the faults of the monitored power transformer.

[0042] First, a cycle of power transformer online monitoring data is collected, and the power transformer online monitoring data is preprocessed, mainly including: data cleaning, data integration, data transformation and data reduction, etc. The power transformer online monitoring data after data preprocessing is loaded into a statistical analysis model, and a statistical rule analysis is performed to determine the distribution statistical model and the model parameters. Load more cycles of power transformer online monitoring data after data preprocessing, input the determined distribution statistical model for further statistical rule analysis, and obtain a plurality of groups of model parameters. Among them, if the power transformer online monitoring data is single parameter, a single parameter distribution statistical model fitting is performed; if the power transformer online monitoring data is multi-parameter, a multi-parameter distribution statistical model fitting is performed. Next, based on the comparison of different model parameters, it is determined whether the power transformer online monitoring data is abnormal, so as to determine the state of the monitored power transformer. Specifically, if the error between the model parameters is within the threshold, it is judged that the power transformer is faulty, otherwise it is judged that the power transformer is faulty.

[0043] Among them, the power transformer online monitoring data at least includes oil dissolved gas data, partial discharge data, core grounding current data, oil temperature, oil pressure, etc. The distribution statistical model includes single parameter statistical model (normal distribution, Weibull distribution, t parameter distribution, etc.), multi-parameter statistical model (multiple linear regression detection, Logistic regression detection, Cox proportional risk regression detection, time series detection).

[0044] The model parameters refer to assuming that the population obeys a certain distribution statistical model, which can be determined by some parameters, such as: mean and standard deviation (normal distribution) and the like. The statistical rule analysis compares the statistical distribution model with the model parameters, including comparing the online monitoring data, the change trend data, and the time series data within a specified time to obtain the online monitoring data distribution, the statistical distribution model and the model parameters, and comparing them to explore the correlation between the online monitoring data and objectively reflect the actual situation.

[0045] The power transformer online monitoring data is obtained by using a rolling time window to extract from real-time online monitoring data. Among them, the rolling time window is a continuous and non-overlapping time window of adjacent two time windows, and the obtained power transformer online monitoring data is at least two sequential time data with the shortest interval time captured in the online monitoring according to the time sequence. The power transformer online monitoring data is obtained and analyzed, including oil dissolved gas, core grounding current, partial discharge, oil temperature, oil pressure, etc.

[0046] The model parameters are obtained by single parameter and multi-parameter regression statistics and analysis of the online monitoring data, and the model parameters are compared to determine whether there is a trend change in the online monitoring data, so as to determine the state of the monitored power transformer.

[0047] The specific process of estimating the state of the monitored power transformer according to single parameter statistics is that in the process of obtaining single parameter data, the single parameter online monitoring data can be directly obtained, such as dissolved gas in oil, partial discharge, core grounding current, oil temperature, oil pressure, etc. Within the same period of time, the operating state of the monitored power transformer is positively correlated with the online monitoring parameter data. By fitting and analyzing the obtained data, the distribution characteristics of the single monitoring parameter can be obtained, such as normal distribution, Weibull distribution and t parameter distribution. According to the distribution characteristics, the operating state of the monitored power transformer is determined.

[0048] The specific process of estimating the state of the monitored power transformer according to multi-parameter regression analysis is that in the process of obtaining two parameters and above monitoring data, within the same period of time, the operating state of the monitored power transformer is correlated with its online monitoring data. By multiple regression linear analysis of the obtained data, the optimal regression equation of different operating conditions can be obtained, and the optimal monitoring quantity that determines the operating condition is determined. According to the distribution characteristics, the operating state of the monitored transformer is determined.

[0049] Further illustrate the single parameter statistical analysis process, take the dissolved gas content in oil as an example, respectively, the time series data of the dissolved gas in oil before and after the partial discharge occurs are statistically analyzed, and the normal distribution, Weibull distribution and t parameter distribution statistics are respectively performed. By comparison, it is found that compared with the normal distribution parameters (0.007555, 0.069186) of the H2 concentration change in oil under normal operating condition, both parameters are increased by several times. It can be considered that an abnormal point has occurred, which has caused a great change in the distribution of data, so as to infer that the power transformer has an abnormal state.

[0050] Further illustrate the multi-state monitoring parameter analysis process, take the dissolved gas, partial discharge and core grounding current as an example, take the dissolved gas in oil as the independent variable Y, and the other monitoring quantities as the dependent variables: partial discharge X1 and core grounding current X2. The multiple regression equation Y = β0 + β1X1 + β2X1 + ε is fitted by using the data, and the optimal equation expression y = 4.429 - 31.009x1 + 32.773x2 is determined. Figure 5 In order to fit the scatter plot, the sub-distribution of each point in the figure has no obvious regularity, and can be considered as approximately random distribution, so the fitting equation can be considered as appropriate. By comparison, it is found that the parameters (β0, β1, β2, ε) of the regression equation have changed greatly. It can be considered that if an abnormal point occurs, the distribution of data will change greatly, so as to infer that the power transformer has an abnormal state.

[0051] Embodiment Two

[0052] The application provides a transformer fault diagnosis system based on statistical characteristics of multiple monitoring variables, comprising:

[0053] A data acquisition and preprocessing module is configured to acquire one cycle of power transformer online monitoring data and perform data preprocessing on the power transformer online monitoring data.

[0054] A model determination module is configured to load a distribution statistical model on the preprocessed power transformer online monitoring data, perform statistical rule analysis based on the loaded statistical distribution model, and determine the distribution statistical model and model parameters.

[0055] A model parameter acquisition module is configured to acquire more cycles of preprocessed power transformer online monitoring data, input the determined distribution statistical model, and obtain a plurality of groups of model parameters.

[0056] A judgment module is configured to compare the model parameters obtained according to different cycles of power transformer online monitoring data, and if the error between the model parameters is within a threshold, it is determined that the power transformer is faulty, otherwise it is determined that the power transformer is faulty.

[0057] Those skilled in the art should understand that the embodiments of the application can be provided as a method, a system, or a computer program product. Therefore, the application can be in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the application can be in the form of a computer program product implemented on one or more computer usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0058] The application is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the computer or other programmable data processing apparatus produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The device specified in one flow or a plurality of flows and / or blocks Figure 1 The device specified in one flow or a plurality of flows and / or blocks

[0059] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the flow Figure 1 of the flow or flows and / or blocks Figure 1 of the block or blocks specified in the flow.

[0060] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flow Figure 1 of the flow or flows and / or blocks Figure 1 of the block or blocks specified in the flow.

[0061] Finally, it should be noted that the above-mentioned embodiments are merely used to illustrate the technical solutions of the present application, but are not intended to limit the protection scope of the present application. Although the present application has been described in detail with reference to the above-mentioned embodiments, it should be understood by those skilled in the art that, after reading the present application, various modifications, replacements or equivalent replacements can be made to the specific embodiments of the present application, but these modifications, replacements or equivalent replacements are all within the protection scope of the claims of the present application.

Claims

1. A transformer fault diagnosis method based on statistical characteristics of multiple monitoring variables, characterized in that, The method comprises the following steps: S100: collecting power transformer online monitoring data of a period, and performing data preprocessing on the power transformer online monitoring data; S200: loading a distribution statistical model on the preprocessed power transformer online monitoring data, and performing statistical rule analysis based on the loaded distribution statistical model to determine the distribution statistical model and model parameters; S300: obtaining preprocessed power transformer online monitoring data of more periods, inputting the determined distribution statistical model to obtain a plurality of groups of model parameters; S400: comparing the model parameters obtained based on the power transformer online monitoring data of different periods, and determining that the power transformer is fault-free if the error between the model parameters is within a threshold, and otherwise determining that the power transformer has a fault.

2. The transformer fault diagnosis method based on statistical characteristics of multiple monitoring quantities according to claim 1, characterized in that, The power transformer online monitoring data comprises oil dissolved gas data, partial discharge data, core grounding current data, oil temperature and oil pressure.

3. The transformer fault diagnosis method based on statistical characteristics of multiple monitoring quantities according to claim 1, characterized in that, The data preprocessing comprises data cleaning, data integration, data transformation and data reduction.

4. The transformer fault diagnosis method based on statistical characteristics of multiple monitoring quantities according to claim 3, characterized in that, The data integration comprises: solving the problem that the time index of the power transformer online monitoring data is not monotonically increasing; solving the problem that the time index of the power transformer online monitoring data contains repeated time stamps by fixing the first value; solving the problem that the power transformer online monitoring data lacks a time index attribute freq; solving the problem that the time series of the power transformer online monitoring data includes a classification label column by converting the classification label into a binary indicator for repair.

5. The transformer fault diagnosis method based on multi-monitoring variable statistical characteristics according to claim 1, characterized in that, The distribution statistical model comprises a single-parameter statistical model and a multi-parameter statistical model.

6. The transformer fault diagnosis method based on multiple monitoring variable statistical characteristics according to claim 5, characterized in that, The single-parameter statistical model comprises a normal distribution, a Weibull distribution and a t-parameter distribution.

7. The transformer fault diagnosis method based on multiple monitoring variable statistical characteristics according to claim 5, characterized in that, The multi-parameter statistical model comprises multiple linear regression detection, Logistic regression detection, Cox proportional risk regression detection and time series detection.

8. The transformer fault diagnosis method based on multi-monitoring variable statistical characteristics according to claim 1, characterized in that, The statistical rule analysis is based on Python language and uses a Pandas package.

9. The transformer fault diagnosis method based on multi-monitoring variable statistical characteristics according to claim 1, characterized in that, The model parameters are parameters capable of determining the distribution statistical model.

10. A transformer fault diagnosis system based on statistical characteristics of multiple monitoring quantities, characterized in that, It comprises: a data acquisition and preprocessing module for collecting power transformer online monitoring data of a period, and performing data preprocessing on the power transformer online monitoring data; a model determination module for loading a distribution statistical model on the preprocessed power transformer online monitoring data, and performing statistical rule analysis based on the loaded distribution statistical model to determine the distribution statistical model and model parameters; a model parameter acquisition module for obtaining preprocessed power transformer online monitoring data of more periods, inputting the determined distribution statistical model to obtain a plurality of groups of model parameters; a judgment module for comparing the model parameters obtained based on the power transformer online monitoring data of different periods, and determining that the power transformer is fault-free if the error between the model parameters is within a threshold, and otherwise determining that the power transformer has a fault.

Images