A method and device for classifying health status of substation equipment

Abstract

The present application discloses a method and device for classifying the health status of substation equipment, the method comprising: constructing an attenuation correlation surface between equipment health and equipment life according to the use time of the substation equipment; based on the attenuation correlation surface, selecting preferred indicators from multiple equipment indicators according to the correlation and the correlation threshold to obtain standard indicators, the equipment indicators include the number of failures, the number of defects, the load rate and the average life; inputting the standard indicators into a preset classification network model for classification processing to obtain equipment status classification results; the preset classification network model is trained by historical standard indicators and corresponding historical equipment status, and the historical equipment status includes historical equipment health and historical equipment life. The present application can solve the technical problem that the prior art lacks intuitive and visual equipment status labels, resulting in limitations in substation equipment status evaluation and lack of accuracy in the results.

Description

Technical Field

[0001] The present application relates to the field of neural network classification technology, and in particular to a method and device for classifying the health status of substation equipment. Background Art

[0002] Monitoring the operating status of substation equipment has always been an important part of daily production in the power system. With the continuous improvement of monitoring technology, the power system has applied a large number of monitoring sensors and equipment to production equipment, built auxiliary monitoring platforms and patrol systems, collected a large amount of data from various aspects such as electrical, appearance, and environment, and visualized the data through various charts. Various places also use various big data analysis methods to evaluate and predict equipment status, discover equipment failures in a timely manner, and conduct daily inspections, maintenance, and emergency repairs, which reduces the impact of equipment failures and improves work efficiency.

[0003] In the prior art, one method is to mine the coupling relationship between multi-dimensional monitoring data and product performance indicators through a deep learning model. Another method is to collect and preprocess the status monitoring signals of the equipment throughout its life cycle, and use wavelet packet analysis, permutation entropy, and information entropy feature screening. Regardless of which method is used, the equipment is directly associated with the indicator, and it is impossible to intuitively reflect the status of the equipment through labels. The selection of indicators is also unreliable, resulting in limitations in the use of each method, which in turn makes the classification results lack accuracy. Summary of the invention

[0004] The present application provides a method and device for classifying the health status of substation equipment, which is used to solve the technical problems that the prior art lacks intuitive equipment status labels, the selection of indicators is unreliable, resulting in limitations in the use of classification methods, and the lack of accuracy in classification results.

[0005] In view of this, the first aspect of the present application provides a method for classifying the health status of a substation, including:

[0006] According to the use time of substation equipment, the attenuation correlation surface between equipment health and equipment life is constructed;

[0007] Based on the attenuation correlation surface, selecting preferred indicators from multiple equipment indicators according to correlation and correlation thresholds to obtain standard indicators, wherein the equipment indicators include the number of failures, the number of defects, the load rate and the average lifespan;

[0008] Input the standard indicators into a preset classification network model for classification processing to obtain a device status classification result;

[0009] The preset classification network model is trained by historical standard indicators and corresponding historical equipment status, and the historical equipment status includes historical equipment health and historical equipment life.

[0010] Preferably, the step of constructing a decay correlation surface between equipment health and equipment life according to the use time of the power transformation equipment includes:

[0011] Draw a curve of equipment failure rate-usage time relationship of the power substation equipment according to the usage time of the power substation equipment;

[0012] Determine the life loss of the power transformation equipment according to the equipment failure rate-usage time relationship curve;

[0013] Based on the life loss, equipment health and equipment age, a decay correlation surface between equipment health and equipment life is drawn.

[0014] Preferably, based on the attenuation correlation surface, selecting a preferred indicator from a plurality of device indicators according to the correlation degree and the correlation threshold to obtain a standard indicator includes:

[0015] Selecting a plurality of equipment indicators based on the attenuation correlation surface;

[0016] Using a preset standardization algorithm to normalize all the equipment indicators to obtain valid indicator data;

[0017] Calculate the correlation between the two valid indicator data by using a preset correlation analysis method;

[0018] The valid indicator data is screened according to the relevance and the relevance threshold to obtain a standard indicator.

[0019] Preferably, the training process of the preset classification network model is:

[0020] Obtain a large number of historical equipment indicators and normalize them to obtain a large number of historical standard indicators;

[0021] Inputting the historical standard indicators and the corresponding historical equipment states into the initial classification network model for training to obtain a training classification network model;

[0022] The training classification network model is tested using a historical indicator test set, and the training classification network model whose test result meets a preset condition is used as a preset classification network model.

[0023] Preferably, the step of inputting the standard indicators into a preset classification network model for classification processing to obtain a device status classification result further includes:

[0024] When the device status classification result does not meet the preset classification standard, the relevant threshold is adjusted and a preset number of the standard indicators are eliminated.

[0025] A second aspect of the present application provides a device for classifying the health status of a substation, including:

[0026] A surface construction module is used to construct a decay correlation surface between equipment health and equipment life according to the use time of the substation equipment;

[0027] An indicator screening module, used to screen preferred indicators from a plurality of equipment indicators based on the attenuation correlation surface according to correlation and correlation threshold, to obtain standard indicators, wherein the equipment indicators include the number of failures, the number of defects, the load rate and the average lifespan;

[0028] A state classification module, used for inputting the standard indicators into a preset classification network model for classification processing to obtain a device state classification result;

[0029] The preset classification network model is trained by historical standard indicators and corresponding historical equipment status, and the historical equipment status includes historical equipment health and historical equipment life.

[0030] Preferably, the curved surface construction module comprises:

[0031] A first drawing submodule is used to draw a device failure rate-usage time relationship curve of the power substation according to the usage time of the power substation;

[0032] A loss determination submodule, used to determine the life loss of the power transformation equipment according to the equipment failure rate-usage time relationship curve;

[0033] The second drawing submodule is used to draw a decay correlation surface between the equipment health and the equipment life based on the life loss, equipment health and equipment age.

[0034] Preferably, the indicator screening module includes:

[0035] An indicator selection submodule, used for selecting a plurality of equipment indicators based on the attenuation correlation surface;

[0036] A normalization processing submodule is used to normalize all the device indicators using a preset standardization algorithm to obtain valid indicator data;

[0037] A correlation calculation submodule, used to calculate the correlation between the effective indicator data in pairs by a preset correlation analysis method;

[0038] The standard screening submodule is used to screen the valid indicator data according to the relevance and the relevant threshold value to obtain the standard indicator.

[0039] Preferably, the training process of the preset classification network model is:

[0040] Obtain a large number of historical equipment indicators and normalize them to obtain a large number of historical standard indicators;

[0041] Inputting the historical standard indicators and the corresponding historical equipment states into the initial classification network model for training to obtain a training classification network model;

[0042] The training classification network model is tested using a historical indicator test set, and the training classification network model whose test result meets a preset condition is used as a preset classification network model.

[0043] Preferably, it also includes:

[0044] The indicator adjustment module is used to adjust the relevant threshold value and eliminate a preset number of standard indicators when the device status classification result does not meet the preset classification standard.

[0045] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:

[0046] In the present application, a method for classifying the health status of substation equipment is provided, including: constructing an attenuation correlation surface between equipment health and equipment life according to the use time of the substation equipment; based on the attenuation correlation surface, selecting preferred indicators from multiple equipment indicators according to correlation and correlation thresholds to obtain standard indicators, and the equipment indicators include the number of failures, the number of defects, the load rate and the average life; inputting the standard indicators into a preset classification network model for classification processing to obtain equipment status classification results; the preset classification network model is trained by historical standard indicators and corresponding historical equipment status, and the historical equipment status includes historical equipment health and historical equipment life.

[0047] The substation equipment health status classification method provided by the present application constructs a correlation surface between equipment health and equipment life, and then selects equipment indicators using the surface relationship, so that there is a correlation relationship between standard indicators and equipment health and equipment life, which can not only ensure the reliability of indicator selection, but also ensure the accuracy of classification results; that is, the present application constructs a label of equipment status composed of equipment health and equipment life, which can more intuitively express the status of the equipment. Therefore, the present application can solve the technical problems that the prior art lacks intuitive equipment status labels, the selection of indicators is unreliable, resulting in limitations in the use of classification methods, and the lack of accuracy in classification results. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A flowchart of a method for classifying the health status of a substation equipment provided in an embodiment of the present application;

[0049] Figure 2Another flowchart of a method for classifying the health status of a substation equipment provided in an embodiment of the present application;

[0050] Figure 3 A schematic diagram of the structure of a device for classifying the health status of a substation provided in an embodiment of the present application;

[0051] Figure 4 A bathtub curve diagram of equipment failure provided by an embodiment of the present application;

[0052] Figure 5 A surface diagram showing the attenuation correlation between the equipment health and equipment life provided in the embodiment of the present application;

[0053] Figure 6 A schematic diagram of the structure of a preset classification network model provided in an embodiment of the present application;

[0054] Figure 7 A schematic diagram of the structure of a preprocessing module in a network model provided in an embodiment of the present application;

[0055] Figure 8 A schematic diagram of the structure of a multi-scale feature extraction module provided in an embodiment of the present application. DETAILED DESCRIPTION

[0056] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0057] For easier understanding, see Figure 1 , Embodiment 1 of a method for classifying the health status of a substation provided by the present application includes:

[0058] Step 101: construct an attenuation correlation surface between equipment health and equipment life according to the use time of the substation equipment.

[0059] The degree of loss of substation equipment with different usage time is different, so the life of the equipment is also different. And the life of equipment can be divided into service life, technical life, depreciation life and economic life. Service life refers to the time from the time the equipment is put into use and operation to the time when it can no longer be used physically or mechanically, or it is naturally scrapped; technical life refers to the total time that the equipment goes through from the time it is put into use and operation to the time when it is finally eliminated due to invisible wear and tear of the equipment caused by technological progress; depreciation life refers to the time it takes to depreciate the remaining value of the equipment to close to zero or the specified residual value in accordance with the regulations of the national financial department; economic life refers to the total time that the equipment goes through from the time it is put into use and operation to the time when it is withdrawn from operation due to excessive operating costs at a certain moment. The comprehensive equipment life can be determined by combining the loss of various related lifespans. This embodiment analyzes the comprehensive equipment life.

[0060] Step 102: Based on the attenuation correlation surface, select preferred indicators from multiple equipment indicators according to correlation and correlation thresholds to obtain standard indicators. The equipment indicators include the number of failures, the number of defects, the load rate and the average lifespan.

[0061] The selection of equipment indicators based on the attenuation correlation surface can ensure that the selected indicators are factors that can affect the health and life of the equipment. That is, not all factors are considered to increase calculation redundancy, nor are independent indicators randomly selected.

[0062] In addition to the number of failures, number of defects, load rate and average life, equipment indicators can also include: the type of region to which the equipment belongs, capacity, power supply, power sales, voltage compliance rate, commissioning date, accumulated depreciation, trouble-free operation time, number of customer complaints and power supply reliability, etc.

[0063] The correlation is the similarity between two indicators, which can be calculated by a specific similarity algorithm, which will not be described in detail. The setting of the correlation threshold can eliminate the indicators with lower correlation and retain the indicators with higher correlation as valid indicators. The valid indicators also need to be unified in dimension before they can become standard indicators for classification tasks. Since the selected indicators can affect the health and life of the equipment, the reliability of the classification results based on the standard indicators can be ensured.

[0064] Step 103: input the standard index into a preset classification network model for classification processing to obtain a device status classification result.

[0065] The preset classification network model is trained by historical standard indicators and corresponding historical equipment status, and the historical equipment status includes historical equipment health and historical equipment life.

[0066] The preset classification network model is a pre-trained classification model. The input in the pre-training process is the historical standard indicator, and the input is the historical equipment status label. It can be understood that when the input is the standard indicator, the input is the equipment status label, that is, the equipment status classification result, and the equipment status label includes equipment health and equipment life.

[0067] The equipment status label database can be constructed based on the health and life of the equipment for easy training and use. Equipment health status: healthy, sub-healthy, mild illness, moderate illness, severe illness, and the equipment health is scored according to different statuses; equipment life status: adolescent, young, middle-aged, middle-aged, and elderly, and the equipment life can be determined according to the usage time. The influencing indicators are selected through the set equipment labels, and then the equipment status is classified through the screened indicators, forming a closed loop of association relationships, ensuring the reliability of the classification results.

[0068] The method for classifying the health status of substation equipment provided in the embodiment of the present application constructs a correlation surface between the health of the equipment and the life of the equipment, and then selects equipment indicators using the surface relationship, so that there is a correlation relationship between the standard indicator and the health of the equipment and the life of the equipment, which can not only ensure the reliability of the indicator selection, but also ensure the accuracy of the classification results; that is, the embodiment of the present application constructs a label of the equipment status composed of the health of the equipment and the life of the equipment, which can more intuitively express the status of the equipment. Therefore, the embodiment of the present application can solve the technical problems that the prior art lacks intuitive equipment status labels, the selection of indicators is unreliable, resulting in limitations in the use of classification methods, and the lack of accuracy in classification results.

[0069] For easier understanding, see Figure 2 The present application provides a second embodiment of a method for classifying the health status of a substation, including:

[0070] Step 201: Draw a curve of equipment failure rate-usage time of the power substation equipment according to the usage time of the power substation equipment.

[0071] See also Figure 4 For different equipment, the number of failures and service life are different, but the development law is the same. Figure 4 It is a curve of the relationship between equipment failure rate and usage time. Its shape is high on both sides and concave and flat in the middle, resembling a bathtub, so it is called the bathtub curve of equipment failure. The bathtub curve divides the life stage of equipment into three stages: early failure period, occasional failure period, and wear and tear failure period.

[0072] Step 202: Determine the life loss of the power transformation equipment according to the equipment failure rate-usage time relationship curve.

[0073] By analyzing the historical operation and maintenance data of substation equipment, the attenuation of equipment life can be formed into a segmented curve. For service life, during the stable operation stage of the equipment, the life decays according to normal time; when a fault occurs in the equipment, the attenuation coefficient is adjusted according to the severity of the fault. For technical life, during the stable operation stage of the equipment, the life decays according to normal time. When the use of equipment is restricted due to technological progress, the attenuation coefficient is adjusted according to the severity of the restriction. For economic life, it decays according to operating costs, and the details are not elaborated here. For depreciation life, it is only related to the number of years.

[0074] Step 203: Draw a decay correlation surface between the equipment health and the equipment life based on the life loss, the equipment health and the equipment age.

[0075] See also Figure 5 According to the life loss, equipment health and equipment age, the attenuation correlation surface between equipment health and equipment life can be drawn, which can describe the correlation between equipment health and equipment life and equipment age.

[0076] Step 204: Select multiple equipment indicators based on the attenuation correlation surface.

[0077] Equipment indicators can be set according to actual conditions, without limiting specific quantities or information categories. As long as they can be quantified, expressed in a unified dimension, and can have an impact on equipment health and equipment life, they can be used as alternative equipment indicators.

[0078] Step 205: normalize all device indicators using a preset standardization algorithm to obtain valid indicator data.

[0079] Different evaluation indicators often have different dimensions and dimensional units (for example: current: A, mA, etc.; voltage: V, mV, etc.), which will affect the results of data correlation analysis. In order to eliminate the impact of the dimensions between indicators, data standardization, or normalization, is required to solve the comparability between data indicators. After the original data is standardized, each indicator is at the same order of magnitude, which is suitable for comprehensive comparative evaluation.

[0080] Min-Max Normalization, also known as deviation normalization, is a linear transformation of the original data, which uniformly maps the result values ​​of different factors to [0, 1]. The conversion function is as follows:

[0081]

[0082] Among them, x is the original data or information, min is the minimum value expression of the current information, and max is the maximum value expression of the current information. Regardless of the type of data information, the above formula can be used for normalized expression and dimension removal.

[0083] Step 206: Calculate the correlation between two valid indicator data by using a preset correlation analysis method.

[0084] The correlation is used to perform factor correlation analysis. The correlation in this embodiment is the correlation coefficient, which is a statistical indicator of the closeness of the relationship between the response variables. The value range of the correlation coefficient is between 1 and -1. 1 means that the two variables are completely linearly correlated, -1 means that the two variables are completely negatively correlated, and 0 means that the two variables are uncorrelated; the closer the data is to 0, the weaker the correlation. The following is the calculation formula of the correlation coefficient:

[0085]

[0086] Among them, S xy is the sample covariance, S x , S y are the sample standard deviations of sample x and sample y respectively. The specific calculation methods of covariance and sample standard deviation are as follows:

[0087]

[0088]

[0089]

[0090] Among them, X i , Y i are the i-th x sample and y sample respectively, is the mean value of sample x, is the mean of sample y, and n is the number of each sample.

[0091] Step 207: Screen the valid indicator data according to the relevance and the relevance threshold to obtain the standard indicator.

[0092] Equipment indicators include the number of failures, number of defects, load rate and average lifespan. In addition, they also include the type of equipment region, capacity, power supply, power sales, voltage qualification rate, commissioning date, accumulated depreciation, trouble-free operation time, number of customer complaints and power supply reliability.

[0093] Multiple correlations r can be calculated xyThe correlation can be screened according to the set correlation threshold. Generally, the indicators with strong correlation, that is, the indicators with larger correlation values, are retained. In this embodiment, the standard indicators finally screened are the number of failures, the number of defects, the load rate and the average lifespan. These indicators are used as evaluation data to classify the health and lifespan of the equipment.

[0094] Step 208: Input the standard index into the preset classification network model for classification processing to obtain the equipment status classification result.

[0095] The preset classification network model is trained by historical standard indicators and corresponding historical equipment status, and the historical equipment status includes historical equipment health and historical equipment life.

[0096] Multiple standard indicators can be converted into vector format and input into the preset classification network model for classification processing to obtain the equipment status classification results, namely equipment health and equipment life.

[0097] The preset classification network model selected in this embodiment is constructed based on a convolutional neural network. The network structure includes a preprocessing module, 8 groups of multi-scale feature extraction modules, a global average pooling layer and a classification output layer; the input layer uses multi-channel one-dimensional data. For the network model structure, please refer to Figure 6 The preprocessing module includes a one-dimensional convolution layer, a normalization layer, an activation layer, and a pooling layer. For the module structure, please refer to Figure 7 The multi-scale feature extraction module is used to extract feature vectors. The extraction of multi-scale features can show that the extracted features fail due to the time offset between the input vector and the device state change; therefore, the feature extraction module includes multiple convolution kernels such as 1X1 convolution, 1x1 convolution to 1x3 convolution, 1x1 convolution to 1x5 convolution, etc. For the module structure, please refer to Figure 8 .

[0098] The damage function selected by the preset classification network model is a comprehensive loss function, which includes a softmax loss function and an intra-class distance minimization loss function; the softmax loss function can maximize the inter-class distance, and calculating the distance between the channel feature and the intra-class center can minimize the intra-class distance; then a weight parameter λ is set to combine the two loss functions to obtain the comprehensive loss function in this embodiment:

[0099]

[0100] Where N is the number of categories, y j is the mark of the sample belonging to the jth class (0 does not belong, 1 does), p j is the confidence that the sample calculated by the softmax layer belongs to the jth class, M is the number of network feature graphs, and s iis the global parity pooling output result of the i-th feature map of the sample, cy i is the center of the i-th feature map of the category to which the sample belongs, represents the Softmax loss function, represents the central loss function and λ is the weight parameter.

[0101] Furthermore, the training process of the preset classification network model is:

[0102] Obtain a large number of historical equipment indicators and normalize them to obtain a large number of historical standard indicators;

[0103] Inputting historical standard indicators and corresponding historical equipment states into the initial classification network model for training to obtain a training classification network model;

[0104] The training classification network model is tested using the historical indicator test set, and the training classification network model whose test results meet the preset conditions is used as the preset classification network model.

[0105] The task of the preset classification network model is to use the input standard indicator vector to perform feature analysis to obtain the equipment status classification result, so the initial structure of the initial classification network model is the same as the structure of the preset classification network model, and the details will not be repeated.

[0106] It can be understood that the screening, processing and operation process of historical equipment indicators are consistent with the relevant operations of standard indicators to maintain the consistency of data expression. In the analysis of test results, the preset conditions and the actual conditions, such as accuracy, recall rate and other model training standards, are not limited here.

[0107] Historical device indicators can be divided into training, validation, and test sets in a ratio of 0.7:0.2:0.1 before pre-training. The training set and validation set are used to pre-train the network model, and the test set is used to test the pre-trained network model.

[0108] Step 209: When the device status classification result does not meet the preset classification standard, adjust the relevant thresholds and eliminate a preset number of standard indicators.

[0109] For the preset classification network model that has been trained, if the device status classification result obtained does not meet the preset classification standard, it can be understood that the accuracy is not high, then the standard indicator needs to be adjusted. The specific adjustment process is to lower the relevant threshold and remove a preset number of standard indicators. The preset number involved can be set according to the actual situation, but the removal standard is to remove a certain number of strongly associated indicators to obtain an updated standard indicator. The updated standard indicator can be reclassified to obtain a new device status classification result; the selection of indicators is constrained by the results to ensure the reliability of the classification results.

[0110] For easier understanding, see Figure 3 The present application also provides an embodiment of a device for classifying the health status of a substation, including:

[0111] A surface construction module 301 is used to construct a decay correlation surface between equipment health and equipment life according to the use time of the substation equipment;

[0112] An indicator screening module 302 is used to screen preferred indicators from multiple equipment indicators based on the attenuation correlation surface, according to the correlation degree and the correlation threshold, to obtain standard indicators, where the equipment indicators include the number of failures, the number of defects, the load rate and the average lifespan;

[0113] The state classification module 303 is used to input the standard index into the preset classification network model for classification processing to obtain the equipment state classification result;

[0114] The preset classification network model is trained by historical standard indicators and corresponding historical equipment status, and the historical equipment status includes historical equipment health and historical equipment life.

[0115] Furthermore, the curved surface construction module 301 includes:

[0116] The first drawing submodule 3011 is used to draw a device failure rate-usage time relationship curve of the power substation equipment according to the usage time of the power substation equipment;

[0117] The loss determination submodule 3012 is used to determine the life loss of the power substation equipment according to the equipment failure rate-use time relationship curve;

[0118] The second drawing submodule 3013 is used to draw a decay correlation surface between the equipment health and the equipment life based on the life loss, the equipment health and the equipment age.

[0119] Furthermore, the indicator screening module 302 includes:

[0120] The indicator selection submodule 3021 is used to select multiple equipment indicators based on the attenuation correlation surface;

[0121] The normalization processing submodule 3022 is used to normalize all device indicators using a preset normalization algorithm to obtain valid indicator data;

[0122] The correlation calculation submodule 3023 is used to calculate the correlation between two valid indicator data by using a preset correlation analysis method;

[0123] The standard screening submodule 3024 is used to screen the valid indicator data according to the relevance and the relevance threshold to obtain the standard indicator.

[0124] Furthermore, the training process of the preset classification network model is:

[0125] Obtain a large number of historical equipment indicators and normalize them to obtain a large number of historical standard indicators;

[0126] Inputting historical standard indicators and corresponding historical equipment states into the initial classification network model for training to obtain a training classification network model;

[0127] The training classification network model is tested using the historical indicator test set, and the training classification network model whose test results meet the preset conditions is used as the preset classification network model.

[0128] Furthermore, it also includes:

[0129] The indicator adjustment module 304 is used to adjust the relevant thresholds and eliminate a preset number of standard indicators when the device status classification result does not meet the preset classification standard.

[0130] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0131] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0132] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0133] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for executing all or part of the steps of the method described in each embodiment of the present application through a computer device (which can be a personal computer, a server, or a network device, etc.). The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (full name in English: Read-Only Memory, English abbreviation: ROM), random access memory (full name in English: Random Access Memory, English abbreviation: RAM), disk or optical disk and other media that can store program codes.

[0134] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for classifying the health status of substation equipment, characterized in that: include: According to the usage time of substation equipment, the attenuation correlation surface between equipment health and equipment life is constructed. The specific process is as follows: Draw a curve of equipment failure rate-usage time relationship of the power substation equipment according to the usage time of the power substation equipment; Determine the life loss of the power transformation equipment according to the equipment failure rate-usage time relationship curve; Draw a decay correlation surface between equipment health and equipment life based on the life loss, equipment health and equipment age; Based on the attenuation correlation surface, the preferred indicators are screened from multiple equipment indicators according to the correlation degree and the correlation threshold to obtain standard indicators. The equipment indicators include the number of failures, the number of defects, the load rate and the average life. The specific process is as follows: Selecting a plurality of equipment indicators based on the attenuation correlation surface; Using a preset standardization algorithm to normalize all the equipment indicators to obtain valid indicator data; Calculate the correlation between the two valid indicator data by using a preset correlation analysis method; Screening the valid indicator data according to the relevance and the relevance threshold to obtain a standard indicator; Input the standard indicators into a preset classification network model for classification processing to obtain a device status classification result; The preset classification network model is trained by historical standard indicators and corresponding historical equipment status, and the historical equipment status includes historical equipment health and historical equipment life.

2. The method for classifying the health status of substation equipment according to claim 1, characterized in that: The training process of the preset classification network model is: Obtain a large number of historical equipment indicators and normalize them to obtain a large number of historical standard indicators; Inputting the historical standard indicators and the corresponding historical equipment states into the initial classification network model for training to obtain a training classification network model; The training classification network model is tested using a historical indicator test set, and the training classification network model whose test result meets a preset condition is used as a preset classification network model.

3. The method for classifying the health status of substation equipment according to claim 1, characterized in that: The step of inputting the standard index into a preset classification network model for classification processing to obtain a device status classification result further includes: When the device status classification result does not meet the preset classification standard, the relevant threshold is adjusted and a preset number of the standard indicators are eliminated.

4. A device for classifying the health status of a substation, characterized in that: include: A surface construction module is used to construct a decay correlation surface between equipment health and equipment life according to the use time of the substation equipment. The surface construction module includes: A first drawing submodule is used to draw a device failure rate-usage time relationship curve of the power substation according to the usage time of the power substation; A loss determination submodule, used to determine the life loss of the power transformation equipment according to the equipment failure rate-usage time relationship curve; A second drawing submodule is used to draw a decay correlation surface between the equipment health and the equipment life based on the life loss, equipment health and equipment age; An indicator screening module is used to screen preferred indicators from multiple equipment indicators based on the attenuation correlation surface according to the correlation degree and the correlation threshold to obtain standard indicators. The equipment indicators include the number of failures, the number of defects, the load rate and the average life. The indicator screening module includes: An indicator selection submodule, used for selecting a plurality of equipment indicators based on the attenuation correlation surface; A normalization processing submodule is used to normalize all the device indicators using a preset standardization algorithm to obtain valid indicator data; A correlation calculation submodule, used to calculate the correlation between the effective indicator data in pairs by a preset correlation analysis method; A standard screening submodule, used for screening the valid indicator data according to the relevance and the relevant threshold value to obtain a standard indicator; A state classification module, used to input the standard indicators into a preset classification network model for classification processing to obtain a device state classification result; The preset classification network model is trained by historical standard indicators and corresponding historical equipment status, and the historical equipment status includes historical equipment health and historical equipment life.

5. The device for classifying the health status of substation equipment according to claim 4, characterized in that: The training process of the preset classification network model is: Obtain a large number of historical equipment indicators and normalize them to obtain a large number of historical standard indicators; Inputting the historical standard indicators and the corresponding historical equipment states into the initial classification network model for training to obtain a training classification network model; The training classification network model is tested using a historical indicator test set, and the training classification network model whose test result meets a preset condition is used as a preset classification network model.

6. The device for classifying the health status of substation equipment according to claim 4, characterized in that: Also includes: The indicator adjustment module is used to adjust the relevant threshold value and eliminate a preset number of standard indicators when the device status classification result does not meet the preset classification standard.

Images