Switch cabinet fault identification method based on supervised learning algorithm and related equipment

Abstract

The invention relates to the technical field of switch cabinet fault identification, in particular to a switch cabinet fault identification method based on a supervised learning algorithm and related equipment. The method comprises the following steps: acquiring a switch cabinet state feature set, optimizing a feature weight corresponding to a sample feature vector by using a weighted KNN algorithm, and recursively dividing samples in the switch cabinet state feature set according to a k value, thereby determining a plurality of nearest neighbors of a to-be-identified sample and corresponding fault mode category labels thereof. And according to the nearest neighbors and the fault mode category labels thereof, the occurrence frequency of each fault category in the nearest neighbors is calculated. And performing majority voting according to the fault type occurrence frequency so as to determine the identification fault type of the switch cabinet. The method is helpful for improving the accuracy and efficiency of fault identification of the switch cabinet.

Description

Technical Field

[0001] This invention relates to the field of switchgear fault identification technology, specifically to a switchgear fault identification method and related equipment based on supervised learning algorithms. Background Technology

[0002] Substation automation systems have undergone a leapfrog development from centralized RTUs to smart substations. Early centralized systems relied on single devices for data acquisition and remote monitoring, but suffered from poor scalability and delayed fault response. Distributed systems improve reliability through modular design, while network-based monitoring systems leverage computer networks for remote operation and maintenance.

[0003] Switchgear, as critical power transmission and distribution equipment, is prone to overheating faults due to abnormal contact resistance (such as oxidation and wear) in its internal contacts / joints. The enclosed structure leads to heat accumulation, and excessively high internal temperatures can cause serious accidents such as insulation failure and fires. Existing diagnostic technologies have limitations: minute changes in contact resistance are difficult to monitor directly, single temperature measurement methods are easily affected by environmental interference, and they lack accurate early warning capabilities for early faults. Therefore, there is an urgent need to develop a multi-parameter fusion online monitoring system that combines data from temperature, current, vibration, and other parameters to achieve early fault identification and location.

[0004] Despite significant progress in substation automation systems and switchgear fault diagnosis technologies, some limitations remain. For example, in substation automation systems, the complexity of data processing and the accuracy of algorithms remain key factors restricting system performance. Simultaneously, with the integration of new energy sources and the development of distributed power generation, the dynamic characteristics of the power grid have become more complex, placing higher demands on the adaptive and collaborative interaction capabilities of substation automation systems. Regarding switchgear fault diagnosis, while existing monitoring technologies can collect data such as contact temperature in real time, they still have limitations in providing early warnings of minute changes in contact resistance and heating phenomena. Furthermore, the development of multi-parameter fusion diagnostic algorithms faces multiple challenges, including data fusion, feature extraction, and pattern recognition. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method and related equipment for identifying switchgear faults based on supervised learning algorithms, which addresses the shortcomings of the existing technology and solves the technical problem of inaccurate fault category identification methods for current switchgear.

[0006] The objective of this invention is achieved through the following technical solutions: In a first aspect, the present invention provides a switchgear fault identification method based on a supervised learning algorithm, comprising: Obtain the switchgear status feature set; the switchgear status feature set includes feature vectors of several samples and corresponding fault mode category labels; The weighted KNN algorithm is used to optimize the feature weights corresponding to the feature vectors of the samples. The samples in the switch cabinet status feature set are recursively divided according to the k value. The nearest neighbors and corresponding fault mode category labels of the divided samples to be identified are determined according to the k value. The frequency of fault categories among the nearest neighbors is calculated based on the nearest neighbors of the sample to be identified and their corresponding fault mode category labels. A majority vote is then conducted based on the frequency of fault categories among the nearest neighbors to determine the fault type of the switchgear.

[0007] As a further improvement of the present invention, the feature weights of the feature vectors are calculated as follows:

[0008] In the formula, The eigenvector of the eigenvector 1 eigenvalue, The eigenvector of the eigenvector 1 eigenvalue, , It is an eigenvalue Feature weights, , Eigenvalues l and eigenvalues j Information gain.

[0009] As a further improvement of the present invention, the step of recursively dividing the samples in the switchgear state feature set according to the k value specifically includes: Obtain the information gain of the eigenvalues ​​in the eigenvector, and select the eigenvalue corresponding to the eigenvector with the largest information gain as the dividing axis; Calculate the median corresponding to all feature values ​​on the dividing axis, and take the samples with feature values ​​less than the median as the left subset and the samples with feature values ​​greater than the median as the right subset; Recursively select the dividing axis and divide the left and right subsets according to the current dividing axis until the number of samples in both the left and right subsets is less than or equal to the value of k.

[0010] As a further improvement of the present invention, it also includes calculating the local density of each sample based on the weighted Euclidean distance, and dynamically adjusting the k value according to the local density of the sample, specifically including: The average distance between each sample in the switch cabinet state feature set is calculated based on the weighted Euclidean distance between any two samples, and the sample neighborhood radius is obtained based on the average distance. The local density of each sample is obtained based on the sample neighborhood radius and weighted Euclidean distance. After normalizing the local density of the samples, the local density dynamics of each sample are calculated based on the normalized local density, and the k value is adjusted based on the local density dynamics of each sample.

[0011] As a further improvement to the present invention, the weighted Euclidean distance between any two samples is:

[0012] In the formula, The weighted Euclidean distance between the feature vectors of samples a and b; and Let a and b represent the feature vectors of samples a and b from any two feature sets in the sample, respectively. , These represent the eigenvectors respectively. and No. One parameter, This represents the parameter weight normalization factor.

[0013] As a further improvement of the present invention, several nearest neighbors and corresponding fault mode category labels of the partitioned sample to be identified are determined based on the k value, specifically including: Recursively compare the feature values ​​of each sample to be identified with the median of the current dividing axis. Samples with feature values ​​less than the median enter the left subtree path, and samples with feature values ​​greater than the median enter the right subtree path. The nearest neighbor search algorithm is used to search for samples in the left subtree path and samples entering the right subtree path to obtain candidate neighbors for the sample to be identified. Select the closest samples among the candidate neighbors as the nearest neighbors.

[0014] As a further improvement of the present invention, the fault categories of the switchgear are as follows:

[0015] In the formula, This refers to the switchgear fault mode category. To identify the fault type for the final switchgear.

[0016] Secondly, the present invention provides a switchgear fault identification system based on a supervised learning algorithm, used to implement the aforementioned switchgear fault identification method based on a supervised learning algorithm, comprising: The data acquisition module acquires a set of switchgear status features; the set of switchgear status features includes feature vectors of several samples and corresponding fault mode category labels; The weighted KNN algorithm execution module optimizes the feature weights corresponding to the feature vectors of the samples using the weighted KNN algorithm, recursively divides the samples in the switch cabinet status feature set according to the k value, and determines several nearest neighbors and corresponding fault mode category labels of the divided samples to be identified according to the k value. The fault type identification module calculates the frequency of fault types among the nearest neighbors of the sample to be identified and their corresponding fault mode category labels. It then uses a majority vote based on the frequency of fault types among the nearest neighbors to determine the fault type of the switchgear.

[0017] Thirdly, the present invention provides a computer-readable storage medium for storing one or more programs, the one or more programs including instructions that, when executed by a computing device, cause the computing device to perform the above-described switchgear fault identification method based on a supervised learning algorithm.

[0018] Fourthly, the present invention provides a computing device, comprising: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include steps for performing the above-described switchgear fault identification method based on supervised learning algorithms.

[0019] The beneficial effects of this invention are as follows: This invention provides a switchgear fault identification method based on a supervised learning algorithm. By using a weighted KNN algorithm, important feature values ​​in the sample feature vectors are assigned higher weights, improving the discriminative power of the feature vectors and thus enhancing the accuracy of fault diagnosis. Through recursive partitioning, the nearest neighbor is selected more precisely, reducing the impact of noise and outliers on the classification results and enhancing the robustness of the system. The majority voting mechanism can effectively suppress the influence of a single erroneous neighbor on the final decision, improving the stability of the system. Through multi-source data fusion and the weighted KNN algorithm, the false diagnosis rate can be effectively reduced, improving the reliability of fault diagnosis. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of the switchgear fault identification method based on supervised learning algorithm in an embodiment of the present invention; Figure 2This is a schematic diagram of the switch cabinet fault identification system based on supervised learning algorithm in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of an electronic device in an embodiment of the present invention. Detailed Implementation

[0022] To make the objectives and technical solutions of this invention clearer and easier to understand, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0023] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings and specific embodiments. The described embodiments are only some embodiments of the present invention, and not all embodiments.

[0024] Example 1 like Figure 1 As shown, the present invention provides a switchgear fault identification method based on a supervised learning algorithm. The following is a specific implementation of an embodiment of the present invention.

[0025] Obtain the switchgear status feature set D, which includes feature vectors of several samples and corresponding fault mode category labels.

[0026] In this embodiment, the set of switchgear state features is as follows: , where m is the total number of samples in the feature set. It contains 19 eigenvalues Feature vector , These are the corresponding fault mode categories, where in this embodiment... When c=1, the switchgear fault mode is normal; when c=2, the switchgear fault mode is excessive bus voltage; when c=3, the switchgear fault mode is excessive cable current; when c=4, the switchgear fault mode is abnormal bus temperature and humidity; when c=5, the switchgear fault mode is excessive temperature at the connection between the bus and the circuit breaker; when c=6, the switchgear fault mode is excessive closing coil current; when c=7, the switchgear fault mode is abnormal circuit breaker temperature and humidity; when c=8, the switchgear fault mode is abnormal instrument temperature and humidity; when c=9, the switchgear fault mode is excessive cable joint temperature; when c=10, the switchgear fault mode is arc fault; when c=11, the switchgear fault mode is system fault; and when c=12, the switchgear fault mode is circuit breaker mechanical fault.

[0027] To ensure optimal classification accuracy across different categories, the weighted KNN algorithm is used to optimize the feature weights corresponding to the feature vectors of the samples. The samples in the switch cabinet status feature set are recursively divided according to the k value, and the nearest neighbors and corresponding fault mode category labels of the divided samples to be identified are determined according to the k value.

[0028] The feature weights of the feature vectors are calculated as follows:

[0029] In the formula, The eigenvector of the eigenvector 1 eigenvalue, The eigenvector of the eigenvector 1 eigenvalue, , It is an eigenvalue Feature weights, , Eigenvalues l and eigenvalues j Information gain.

[0030] The samples in the switchgear state feature set are recursively divided according to the value of k, specifically including: Obtain the information gain of the eigenvalues ​​in the eigenvector, and select the eigenvalue corresponding to the eigenvector with the largest information gain as the dividing axis; Calculate the median of all eigenvalues ​​on the dividing axis. Samples with eigenvalues ​​less than the median are designated as the left subset. Samples with eigenvalues ​​greater than the median are designated as the right subset. ; Recursively select the dividing axis and divide the left and right subsets according to the current dividing axis until the number of samples in both the left and right subsets is less than or equal to the value of k.

[0031] Specifically, to accelerate the efficiency of switchgear fault identification, the feature with the largest information gain is selected as the dividing axis x, and the calculation formula is as follows:

[0032] in, The eigenvector of the eigenvector The eigenvalues ​​of .

[0033] In addition, this embodiment also includes calculating the local density of each sample based on the weighted Euclidean distance, and dynamically adjusting the k value based on the local density of the samples, specifically including: The average distance between each sample in the switch cabinet state feature set is calculated based on the weighted Euclidean distance between any two samples, and the sample neighborhood radius is obtained based on the average distance. The local density of each sample is obtained based on the sample neighborhood radius and weighted Euclidean distance. After normalizing the local density of the samples, the local density dynamics of each sample are calculated based on the normalized local density, and the k value is adjusted based on the local density dynamics of each sample.

[0034] Specifically, the weighted Euclidean distance between the feature vectors of any two feature sets, samples a and b, in the input feature set. The calculation formula is as follows:

[0035] In the formula, The weighted Euclidean distance between the feature vectors of samples a and b; and Let a and b represent the feature vectors of samples a and b from any two feature sets in the sample, respectively. , These represent the eigenvectors respectively. and No. One parameter, This represents the parameter weight normalization factor.

[0036] To improve recognition performance, a dynamically selected k-value is used, which is the average distance between samples in the feature set. The calculation formula is:

[0037] Where m is the total number of samples in the feature set. The weighted Euclidean distance between samples a and b of any two feature sets in the input feature set. The formula for calculating the sample neighborhood radius r of the feature set is: .

[0038] The value of k is dynamically adjusted based on the local density of samples in the feature set, where the local density of sample i in the feature set is... The calculation formula is:

[0039] in, The indicator function represents a sample of the feature set. Whether it is in the feature set of samples Within the radius r of the domain, the value is 1 if it is within the domain, and 0 if it is not. This is the weighted Euclidean distance between samples a and b of any two feature sets in the input feature set.

[0040] Normalized local density The calculation formula is:

[0041] in, Let be the local density maximum value among samples of a set of m features. Let be the local density minimum among samples of m feature sets; Based on the local density dynamics of sample i in the feature set The calculation formula is:

[0042] in, Let i be the local density of sample i in the normalized feature set. and These represent the maximum and minimum values ​​of k. This is verified experimentally. and Time-based classification yields the best results, therefore we choose... and .

[0043] Based on the k-value, determine several nearest neighbors of the sample to be identified after partitioning and their corresponding fault mode category labels, specifically including: Recursively compare the feature values ​​of each sample to be identified with the median of the current dividing axis. Samples with feature values ​​less than the median enter the left subtree path, and samples with feature values ​​greater than the median enter the right subtree path. The nearest neighbor search algorithm is used to search for samples in the left subtree path and samples entering the right subtree path to obtain candidate neighbors for the sample to be identified. Select the closest samples among the candidate neighbors as the nearest neighbors, and finally select k neighbors.

[0044] The frequency of fault categories among the nearest neighbors is calculated based on the nearest neighbors of the sample to be identified and their corresponding fault mode category labels. A majority vote is then conducted based on the frequency of fault categories among the nearest neighbors to determine the fault type of the switchgear.

[0045] After selecting the k nearest neighbors, count the number of category labels to which the neighbors belong, and calculate the frequency of each category c. The calculation formula is:

[0046] in, It is an indicator function, if The value is 1 if it is 1, otherwise it is 0. This refers to the switchgear failure mode category. The instance to be classified is determined by majority voting based on the frequency of occurrence of the k neighboring categories. Ultimately, the fault category of the switchgear is:

[0047] In the formula, This refers to the switchgear fault mode category. To identify the fault type for the final switchgear.

[0048] Example 2 This embodiment is a further improvement on embodiment 1, and the feature signal is designed as follows: Switchgear can be divided into indoor switchgear and outdoor switchgear according to its usage environment. This paper focuses on indoor switchgear. Compared to outdoor switchgear, the working environment of indoor switchgear is less harsh, and the requirements for the working environment of sensors are not as strict. The main structure of the switchgear includes the cabinet, busbar compartment, circuit breaker compartment, cable compartment, and instrument compartment.

[0049] The status characteristic parameters of switchgear condition monitoring should reflect the current operating status of the switchgear from different perspectives. Therefore, the extraction of status characteristic parameters from the monitoring signals is crucial for switchgear fault diagnosis. Starting from the incoming end of the switchgear, monitoring targets are selected sequentially in the busbar compartment, circuit breaker compartment, and cable compartment. Monitoring targets are also set up in the instrument room. The characteristics representing the switchgear status are divided into electrical parameters (voltage, current, reactive power, active power, etc.) and non-electrical parameters (temperature, humidity, flash signal). The sensor locations are distributed as follows: Figure 2 As shown, the specific explanation is as follows: Busbar compartment: 3 types of sensors: Sensor 1: Voltage transformer to collect busbar voltage; Sensor 2: Fiber optic temperature sensor to collect temperature at the connection between the busbar and the circuit breaker; Sensor 3: Wireless temperature and humidity sensor to collect ambient temperature and humidity of the busbar compartment.

[0050] Circuit breaker compartment: 3 types of sensors. Sensor 1 is an optical fiber temperature sensor that collects the temperature of the circuit breaker contacts; sensor 2 is a wireless temperature and humidity sensor that collects the temperature and humidity of the circuit breaker room; sensor 3 is a current transformer that collects the current of the circuit breaker opening and closing coils and the current of the circuit breaker breaking.

[0051] Cable compartment: 3 types of sensors. Sensor 1 is a current transformer that collects the cable outgoing current, which together with the bus voltage obtains the active power, reactive power and power factor; Sensor 2 is an optical fiber temperature sensor that collects the temperature of the cable joint; Sensor 3 is a wireless temperature and humidity sensor that collects the room temperature and humidity of the cable compartment.

[0052] Instrument Room: Wireless temperature and humidity sensors that collect data on the temperature and humidity of the instrument room.

[0053] The characteristic parameters for switchgear status diagnosis are shown in Table 1.

[0054]

[0055] To address the nonlinear classification problem of switchgear status identification, this paper proposes a status determination model based on a k-nn network. By extracting state feature parameters from the monitoring signal and inputting them into the k-nn network for information processing, the automatic determination of the switchgear's operating status can be achieved. The correspondence between the state feature parameters and preset fault identification items is shown in Table 2. The specific implementation steps are as follows: 1) Selecting state feature parameters from the switchgear status sensor monitoring signals as the model input, the calculated state feature parameters should be normalized before being input into the constructed diagnostic network model. Different state feature parameters often have different dimensions, which can affect the output of k-nn. To eliminate the influence of dimensions between state feature parameters, data standardization is required to ensure comparability between data indicators. Studies have shown that after normalization, the state feature parameters are on the same order of magnitude, which can accelerate the gradient descent to find the optimal solution. This paper normalizes the calculated state feature parameters to the range [0,1], as shown in the following expression:

[0056] in , , These represent the i-th data point, minimum value, and maximum value of the same state feature parameter, respectively. This is the result of normalization.

[0057] 2) Divide the sample data into training samples and test samples according to a certain ratio.

[0058] 3) Use the optimal parameters of the training samples to test the test samples.

[0059]

[0060] 4) Define the risk assessment indicators for switchgear failures, and let the risk assessment indicators for switchgear failures... Represented as: ,in This indicates that the switchgear has a fault mode. This indicates the severity of the switchgear malfunction. This indicates the number of scheduled maintenance operations for the switchgear. The risk value of a switchgear failure indicates the extent of damage caused by such a failure. It can be represented as:

[0061] in, , , , These represent the weights of each evaluation indicator, and Each weight can be adjusted according to specific circumstances; the default value is [value missing]. The definitions of each attribute are as follows:

[0062]

[0063]

[0064]

[0065] 6) Based on the risk score of switchgear failures, the priority of switchgear failure repairs can be determined, providing a reference for subsequent maintenance by staff. The specific classification method is as follows: .

[0066] Example 3 like Figure 2 As shown, this embodiment provides a switchgear fault identification system based on a supervised learning algorithm, used to implement the switchgear fault identification method based on a supervised learning algorithm in Embodiment 1. The system includes: The data acquisition module acquires a set of switchgear status features; the set of switchgear status features includes feature vectors of several samples and corresponding fault mode category labels; The weighted KNN algorithm execution module optimizes the feature weights corresponding to the feature vectors of the samples using the weighted KNN algorithm, recursively divides the samples in the switch cabinet status feature set according to the k value, and determines several nearest neighbors and corresponding fault mode category labels of the divided samples to be identified according to the k value. The fault type identification module calculates the frequency of fault types among the nearest neighbors of the sample to be identified and their corresponding fault mode category labels. It then uses a majority vote based on the frequency of fault types among the nearest neighbors to determine the fault type of the switchgear.

[0067] Example 4 In another embodiment of the present invention, a computer-readable storage medium is provided as a storage component within a terminal device, the function of which is to store programs and data. It should be noted that the computer-readable storage medium here encompasses not only the built-in storage components of the terminal device but also extended storage components supported by the device. Essentially, it is a tangible medium capable of containing or storing programs that can be invoked by or in conjunction with an instruction execution system, device, or apparatus. This storage medium provides storage areas for the terminal's operating system and stores one or more instructions suitable for processor loading and execution, which can constitute one or more computer programs containing program code.

[0068] Specifically, examples of computer-readable storage media (a non-exclusive list) include: electrical connections with one or more wires, portable disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, optical fibers, portable optical disc read-only memory, optical storage devices, magnetic storage devices, or any reasonable combination of the above types.

[0069] The storage medium may also include data signals propagated as part of a baseband portion or a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any reasonable combination of both. Furthermore, computer-readable storage medium may also refer to other readable media besides conventional readable storage media, capable of sending, propagating, or transmitting programs for use or operation by an instruction execution system, apparatus, or device. Program code on the storage medium can be transmitted via any suitable medium, including but not limited to wireless, wired, optical fiber, or any reasonable combination thereof.

[0070] The program code used to implement the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C. The execution modes of the program code include: running entirely on the user's computing device, running partially on the user's device as a standalone software package, running partially in a distributed manner on both the user's device and a remote computing device, or running entirely on a remote computing device or server. When a remote computing device is involved, the device can be connected to the user's computing device via any type of network such as a local area network (LAN) or a wide area network (WAN), or connected to an external computing device via the Internet through an Internet service provider.

[0071] The processor is capable of loading and executing one or more instructions stored in a computer-readable storage medium to implement the corresponding steps of the switchgear fault identification method based on supervised learning algorithm described in Example 1.

[0072] Example 5 Figure 3 This is a schematic diagram of a computer device provided according to an embodiment of the present invention.

[0073] Please see Figure 3The terminal device is a computer device. In this embodiment, the computer device 60 includes a processor 61, a memory 62, and a computer program 63 stored in the memory 62 and executable on the processor 61. When the processor 61 executes the computer program 63, it implements the switchgear fault identification method based on the supervised learning algorithm described in this embodiment. To avoid repetition, these details are not elaborated here. Alternatively, when the processor 61 executes the computer program 63, it implements the functions of each model / unit in the computer system that constitutes the switchgear fault identification process based on the supervised learning algorithm described in this embodiment. To avoid repetition, these details are not elaborated here.

[0074] Computer device 60 can be a desktop computer, laptop, handheld computer, cloud server, or other computing device. Computer device 60 may include, but is not limited to, a processor 61 and a memory 62. Those skilled in the art will understand that... Figure 3 This is merely an example of computer device 60 and does not constitute a limitation on computer device 60. It may include more or fewer components than shown, or combine certain components, or different components. For example, computer device may also include input / output devices, network access devices, buses, etc.

[0075] The processor 61 may be a central processing unit (CPU), or other general-purpose processors, CPUs, graphics processing units (GPUs), digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, quantum computing-based data processing logic units, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0076] The memory 62 can be an internal storage unit of the computer device 60, such as a hard disk or RAM of the computer device 60. The memory 62 can also be an external storage device of the computer device 60, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. equipped on the computer device 60.

[0077] Furthermore, the memory 62 may include both internal storage units of the computer device 60 and external storage devices. The memory 62 is used to store computer programs and other programs and data required by the computer device. The memory 62 can also be used to temporarily store data that has been output or will be output.

[0078] Any references to memory, databases, or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0079] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

Claims

1. A method for identifying switchgear faults based on a supervised learning algorithm, characterized in that, include: Obtain the switchgear status feature set; the switchgear status feature set includes feature vectors of several samples and corresponding fault mode category labels; The weighted KNN algorithm is used to optimize the feature weights corresponding to the feature vectors of the samples. The samples in the switch cabinet status feature set are recursively divided according to the k value. The nearest neighbors and corresponding fault mode category labels of the divided samples to be identified are determined according to the k value. The frequency of fault categories among the nearest neighbors is calculated based on the nearest neighbors of the sample to be identified and their corresponding fault mode category labels. A majority vote is then conducted based on the frequency of fault categories among the nearest neighbors to determine the fault type of the switchgear.

2. The switchgear fault identification method based on supervised learning algorithm according to claim 1, characterized in that, The feature weights of the feature vectors are calculated as follows: In the formula, The eigenvector of the eigenvector 1 eigenvalue, The eigenvector of the eigenvector 1 eigenvalue, , It is an eigenvalue Feature weights, , Eigenvalues l and eigenvalues j Information gain.

3. The switchgear fault identification method based on supervised learning algorithm according to claim 1, characterized in that, The recursive partitioning of the switchgear state feature set based on the k value specifically includes: Obtain the information gain of the eigenvalues ​​in the eigenvector, and select the eigenvalue corresponding to the eigenvector with the largest information gain as the dividing axis; Calculate the median corresponding to all feature values ​​on the dividing axis, and take the samples with feature values ​​less than the median as the left subset and the samples with feature values ​​greater than the median as the right subset; Recursively select the dividing axis and divide the left and right subsets according to the current dividing axis until the number of samples in both the left and right subsets is less than or equal to the value of k.

4. The switchgear fault identification method based on supervised learning algorithm according to claim 3, characterized in that, It also includes calculating the local density of each sample based on the weighted Euclidean distance, and dynamically adjusting the k value based on the local density of the samples, specifically including: The average distance between each sample in the switch cabinet state feature set is calculated based on the weighted Euclidean distance between any two samples, and the sample neighborhood radius is obtained based on the average distance. The local density of each sample is obtained based on the sample neighborhood radius and weighted Euclidean distance. After normalizing the local density of the samples, the local density dynamics of each sample are calculated based on the normalized local density, and the k value is adjusted based on the local density dynamics of each sample.

5. The switchgear fault identification method based on supervised learning algorithm according to claim 4, characterized in that, The weighted Euclidean distance between any two samples is: In the formula, The weighted Euclidean distance between the feature vectors of samples a and b; and Let a and b represent the feature vectors of samples a and b from any two feature sets in the sample, respectively. , These represent the eigenvectors respectively. and No. One parameter, This represents the parameter weight normalization factor.

6. The switchgear fault identification method based on supervised learning algorithm according to claim 4, characterized in that, Based on the k-value, determine several nearest neighbors of the sample to be identified after partitioning and their corresponding fault mode category labels, specifically including: Recursively compare the feature values ​​of each sample to be identified with the median of the current dividing axis. Samples with feature values ​​less than the median enter the left subtree path, and samples with feature values ​​greater than the median enter the right subtree path. The nearest neighbor search algorithm is used to search for samples in the left subtree path and samples entering the right subtree path to obtain candidate neighbors for the sample to be identified. Select the closest samples among the candidate neighbors as the nearest neighbors.

7. The switchgear fault identification method based on supervised learning algorithm according to claim 1, characterized in that, The fault categories of the switchgear are: In the formula, This refers to the switchgear fault mode category. To identify the fault type for the final switchgear.

8. A switchgear fault identification system based on a supervised learning algorithm, used to implement the switchgear fault identification method based on a supervised learning algorithm as described in any one of claims 1 to 7, characterized in that, include: The data acquisition module acquires a set of switchgear status features; the set of switchgear status features includes feature vectors of several samples and corresponding fault mode category labels; The weighted KNN algorithm execution module optimizes the feature weights corresponding to the feature vectors of the samples using the weighted KNN algorithm, recursively divides the samples in the switch cabinet status feature set according to the k value, and determines several nearest neighbors and corresponding fault mode category labels of the divided samples to be identified according to the k value. The fault type identification module calculates the frequency of fault types among the nearest neighbors of the sample to be identified and their corresponding fault mode category labels. It then uses a majority vote based on the frequency of fault types among the nearest neighbors to determine the fault type of the switchgear.

9. A computer-readable storage medium for storing one or more programs, characterized in that, The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform the switch cabinet fault identification method based on any one of claims 1 to 7.

10. A computing device, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including steps for performing the switch cabinet fault identification based on any one of claims 1 to 7.

Images