Air switch cabinet discharge detection method and device based on QPSO-SVM

Abstract

The invention provides an air switch cabinet discharge detection method and device based on a QPSO-SVM (Quantum Particle Swarm Optimization-Support Vector Machine), so as to improve the accuracy of fault detection. The method comprises the following steps: S1, acquiring multi-parameter monitoring data of a plurality of air switch cabinets and fault tags corresponding to the multi-parameter monitoring data to form a training sample set, and collecting multi-parameter monitoring data of the air switch cabinets to be tested to form a test data set; s2, a QPSO-SVM model is constructed, a particle swarm of a QPSO algorithm is initialized, the particle swarm comprises N particles, the position of each particle adopts quantum bit coding, the quantum bit code corresponding to each particle represents a group of candidate values of a penalty parameter and a kernel function parameter of the SVM, and N is a positive integer larger than zero; s3, inputting the multi-parameter monitoring data of each sample in the training samples into the QPSO-SVM model for training, and obtaining an optimized QPSO-SVM model; and S4, inputting multi-parameter monitoring data of the air switch cabinet to be detected into the optimized SVM model for prediction, and outputting a corresponding fault label.

Description

Technical Field

[0001] This application relates to the field of non-contact condition detection technology for vacuum circuit breakers, specifically to a discharge detection method and device for air switchgear based on QPSO-SVM. Background Technology

[0002] As a common power transmission and distribution device, the operational reliability of air switchgear directly affects the power supply quality and safety performance of power systems. It is mainly used in various large substations and distribution rooms, and is a crucial component of power systems. However, under the influence of factors such as system overvoltage, insulation defects, manufacturing quality, and operating environment, the electric field strength of certain gases near the conductors in the air switchgear may reach the breakdown level, causing ionization and excitation of the gases, resulting in partial discharge. This damages the insulation of the electrical equipment, further exacerbates the uneven electric field distribution, and intensifies the discharge, creating a vicious cycle. Ultimately, this leads to electrical aging or insulation breakdown of the equipment, resulting in decreased insulation performance, shortened equipment lifespan, and in severe cases, short circuits and power grid safety accidents.

[0003] Practice has proven that discharge phenomena caused by insulation deterioration are one of the early symptoms of insulation problems in electrical equipment and an important indicator for judging the severity of defects. Currently, commonly used discharge detection methods include pulse current method, ultra-high frequency method, ultrasonic method, and transient voltage-to-ground method. These methods all have certain detection capabilities, but they are basically limited to manual inspection or require the placement of detection cabinets near the equipment, resulting in high costs and unsuitability for monitoring main power distribution equipment. Furthermore, most are detection methods used in the middle to late stages of fault occurrence, failing to provide timely online monitoring of discharge phenomena and thus unable to effectively control faults in their early stages.

[0004] Therefore, it is necessary to propose a detection method that is suitable for monitoring main power distribution equipment, can be connected to online monitoring devices, and has a high recognition rate. Summary of the Invention

[0005] This application provides a discharge detection method and device for air switchgear based on QPSO-SVM, which can effectively improve detection accuracy, reduce fault risk, and optimize maintenance strategies, thereby enhancing the operational reliability and safety of air switchgear.

[0006] To achieve the above objectives, this application adopts the following technical solution: Firstly, this application provides a discharge detection method for air switchgear based on QPSO-SVM, the method comprising: S1. Acquire multi-parameter monitoring data and corresponding fault labels of multiple air switch cabinets to form a training sample set, and collect multi-parameter monitoring data of the air switch cabinet under test to form a test dataset. S2, construct the QPSO-SVM model and initialize the particle swarm of the QPSO algorithm. The particle swarm consists of N particles, and the position of each particle is encoded by qubits. The qubit code corresponding to each particle represents a set of candidate values ​​for the penalty parameter and kernel function parameter of the SVM. N is a positive integer greater than zero. S3, input the multi-parameter monitoring data of each sample in the training sample set into the QPSO-SVM model for training, to obtain the optimized QPSO-SVM model, specifically including: Training the QPSO-SVM model involves performing the following steps for each particle in the particle swarm: Step 1: Based on the qubit encoding corresponding to the i-th particle, decode to obtain the specific SVM penalty parameters and kernel function parameters, configure the SVM, and obtain the configured SVM, where i is a positive integer greater than zero and less than or equal to N; Step 2: Input the multi-parameter monitoring data of each sample in the training samples into the configured SVM, and map the multi-parameter monitoring data to a high-dimensional feature space through the kernel function; Step 3: Solve for the optimal classification hyperplane in the high-dimensional feature space and construct the classification model; Step 4: Based on the classification accuracy of the classification model for each sample in the training samples, calculate the fitness value of the i-th particle, where the fitness value is used to characterize the classification accuracy of the configured SVM for the fault labels corresponding to the multi-parameter monitoring data. Step 5: Repeat steps 1 to 4 to obtain the fitness value for each particle; Step 6: Based on the fitness value of each particle, update the individual historical best position of each particle and the global best position of the particle swarm. Step 7: Based on the individual historical best position of each particle and the global best position of the particle swarm, adjust the penalty parameters and kernel function of the SVM, then reconfigure the SVM model, and repeat steps 1 to 6 until the preset maximum number of iterations is reached to obtain the optimized SVM model. S4 inputs the multi-parameter monitoring data of the air switch cabinet under test into the optimized SVM model for prediction and outputs the corresponding fault label.

[0007] Secondly, this application provides a discharge detection device for air switchgear based on QPSO-SVM, the device comprising: The data acquisition module is used to acquire multi-parameter monitoring data and corresponding fault labels of multiple air switch cabinets to form a training sample set, and to acquire multi-parameter monitoring data of the air switch cabinet under test to form a test dataset. The host computer is used to construct the QPSO-SVM model and initialize the particle swarm of the QPSO algorithm. The particle swarm consists of N particles, each with a position encoded using qubits. The qubit encoding for each particle represents a set of candidate values ​​for the SVM's penalty parameters and kernel function parameters, where N is a positive integer greater than zero. It is also used to input the multi-parameter monitoring data of each sample in the training sample set into the QPSO-SVM model for training, obtaining an optimized QPSO-SVM model. Specifically, training the QPSO-SVM model involves performing the following steps for each particle in the particle swarm: Step 1: Based on the qubit encoding corresponding to the i-th particle, decode the specific SVM penalty parameters and kernel function parameters, configure the SVM, and obtain the configured SVM, where i is a positive integer greater than zero and less than or equal to N; Step 2: Input the multi-parameter monitoring data of each sample in the training sample set into the training sample set. The process involves several steps: First, the configured SVM maps the multi-parameter monitoring data to a high-dimensional feature space using a kernel function. Second, the optimal classification hyperplane is solved in the high-dimensional feature space to construct a classification model. Third, based on the classification accuracy of the model for each sample in the training samples, the fitness value of the i-th particle is calculated. The fitness value characterizes the classification accuracy of the configured SVM for the fault labels corresponding to the multi-parameter monitoring data. Fourth, steps one through four are repeated to obtain the fitness value of each particle. Fifth, based on the fitness value of each particle, the individual historical best position and the global best position of the particle swarm are updated. Sixth, based on the individual historical best position and the global best position of the particle swarm, the penalty parameters and kernel function of the SVM are adjusted, and the SVM model is reconfigured. Steps one through six are repeated until the preset maximum number of iterations is reached to obtain the optimized SVM model. The host computer is also used to input the multi-parameter monitoring data of the air switch cabinet under test into the optimized SVM model for prediction and output the corresponding fault labels.

[0008] Thirdly, a computer-readable storage medium is provided, the computer-readable storage medium including storage of a computer program or instructions, which, when executed, cause the QPSO-SVM-based air switchgear discharge detection method of the first aspect to be performed.

[0009] In this application embodiment, by using multi-parameter monitoring data, including various data related to discharge and defects, the status of the air switchgear can be analyzed from multiple dimensions. By using an improved QPSO (Quantum Particle Swarm Optimization) algorithm to optimize the penalty parameters and kernel function parameters of the SVM (Support Vector Machine) model, the accuracy of fault detection can be greatly improved, effectively avoiding the local optimum problem in traditional algorithms and finding the global optimum solution. This ensures that the SVM model can accurately classify discharge type, discharge degree, defect type, and defect degree. In summary, this application can effectively improve detection accuracy, reduce fault risk, and optimize maintenance strategies, thereby enhancing the operational reliability and safety of the air switchgear, reducing economic losses caused by equipment failure, and improving the stability and efficiency of the entire power system.

[0010] Other features and advantages of this application will be described in detail in the following detailed description section. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A schematic flowchart of the discharge detection method for air switchgear based on QPSO-SVM provided in the embodiments of this application; Figure 2 This is a schematic diagram of the structure of the air switch cabinet discharge detection device based on QPSO-SVM provided in the embodiments of this application. Detailed Implementation

[0013] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. At the same time, in the description of the embodiments of this application, the terms "first," "second," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0014] Figure 1This is a flowchart illustrating the discharge detection method for air switchgear based on QPSO-SVM provided in the embodiments of this application.

[0015] The procedure for discharge detection in air switchgear based on QPSO-SVM is as follows: Step S1: Obtain multi-parameter monitoring data and corresponding fault labels for multiple air switch cabinets to form a training sample set, and collect multi-parameter monitoring data of the air switch cabinet under test to form a test dataset. Step S2: Construct the QPSO-SVM model and initialize the particle swarm of the QPSO algorithm. The particle swarm consists of N particles, and the position of each particle is encoded by qubits. The qubit code corresponding to each particle represents a set of candidate values ​​for the penalty parameter and kernel function parameter of the SVM. N is a positive integer greater than zero. Step S3 involves inputting the multi-parameter monitoring data of each sample in the training sample set into the QPSO-SVM model for training, resulting in the optimized QPSO-SVM model. This specifically includes: Training the QPSO-SVM model involves performing the following steps for each particle in the particle swarm: Step 1: Based on the qubit encoding corresponding to the i-th particle, decode to obtain the specific SVM penalty parameters and kernel function parameters, configure the SVM, and obtain the configured SVM, where i is a positive integer greater than zero and less than or equal to N; Step 2: Input the multi-parameter monitoring data of each sample in the training samples into the configured SVM, and map the multi-parameter monitoring data to a high-dimensional feature space through the kernel function; Step 3: Solve for the optimal classification hyperplane in the high-dimensional feature space and construct the classification model; Step 4: Based on the classification accuracy of the classification model for each sample in the training samples, calculate the fitness value of the i-th particle, where the fitness value is used to characterize the classification accuracy of the configured SVM for the fault labels corresponding to the multi-parameter monitoring data. Step 5: Repeat steps 1 to 4 to obtain the fitness value for each particle; Step 6: Based on the fitness value of each particle, update the individual historical best position of each particle and the global best position of the particle swarm. Step 7: Based on the individual historical best position of each particle and the global best position of the particle swarm, adjust the penalty parameters and kernel function of the SVM, then reconfigure the SVM model, and repeat steps 1 to 6 until the preset maximum number of iterations is reached to obtain the optimized SVM model. Step S4: Input the multi-parameter monitoring data of the air switch cabinet under test into the optimized SVM model for prediction and output the corresponding fault label.

[0016] In the QPSO-SVM (Quantum Particle Swarm Optimization-Support Vector Machine) model, the fitness value is an indicator used to evaluate the performance of the model corresponding to each particle (i.e., each combination of SVM parameters). Its core function is to guide the QPSO algorithm to search in the direction of optimizing SVM classification performance.

[0017] In the QPSO-SVM (Quantum Particle Swarm Optimization-Support Vector Machine) model, the fitness value is an indicator that evaluates the performance of the model corresponding to each particle (i.e., each set of SVM parameters) in the QPSO algorithm. It is calculated as follows: the SVM model is configured using the parameters represented by that particle, and then tested on training samples with known fault labels. The model's predictions are compared with the actual fault labels, typically using classification accuracy as the specific fitness value. Therefore, the fitness value directly quantifies the model's ability to identify and match real fault patterns under the current parameters; its level depends entirely on the degree of agreement between the model's predictions and the fault labels.

[0018] The position of each of the above particles is encoded using qubits, specifically for each particle. The position is determined by a set of quantum vertices and This indicates that it constitutes a position code. ,in and Related to the penalty parameter and kernel function parameters , rand is a random number in the range [0,1]. ; ; m Population size; n To solve for the spatial dimension, in this problem, n That is, 2, when j When =1, it corresponds to the penalty parameter, when j When =2, it corresponds to the kernel parameter; and Corresponding to the positions of cosine and sine.

[0019] It should also be noted that in the improved QPSO, a particle simultaneously occupies two positions in the solution space. The following formula transforms the two unit spatial positions of the particle into sinusoidal positions in the solution space of the optimization problem. Sum and cosine position .

[0020] ; in, 、 The first and second particles respectively jThe position (for the optimization problem) j The minimum and maximum values ​​of (variables), when j When =1, and This corresponds to the two values ​​of the penalty parameter in the solution space, when j When =2, and These correspond to the two values ​​of the kernel parameter in the solution space.

[0021] The position of the particle is updated using the wave function equation based on quantum behavior properties, and the result is returned. The value constitutes the updated particles. i The current position is encoded. Updating only the position variable greatly improves the algorithm's search capability and convergence speed.

[0022] Regarding the first k Iteration, for particles i Solution space sine position and cosine position The update is performed based on the fitness function value of the particle solution space (which is the corresponding penalty parameter). and nuclear parameters and The accuracy of SVM in identifying discharge type, defect type, and severity classification based on multi-parameter detection data was determined. and A better position The position is updated to a better location, resulting in the updated position. Then return The value of , thus constituting a newer particle. i Current position code The location update process is as follows: ; ; ; ; ; ; in, , These represent the optimal position of an individual particle and the optimal position of the entire population, respectively. , u All are random numbers in the range [0,1]. For particles i No. k The local attraction region of the next iteration; It is the first k The average of the optimal positions of all individual particles in the next iteration population; L i ( k ) indicates the first k Sub-iteration particles i The weighted distance from the population's average optimal position; It is a contraction-expansion factor, which typically decreases linearly; a , b These correspond to the minimum and maximum values ​​in the particle solution space, respectively.

[0023] In this embodiment, the multi-parameter monitoring data includes data on ozone, carbon monoxide, nitrogen dioxide, ultraviolet pulse, temperature and humidity, and smoke particle concentration. The fault label includes at least the discharge type, discharge degree, defect type, and defect degree.

[0024] The discharge types include corona, arc, and surface flashover, and the discharge degree includes mild, moderate, and severe. The defect types include insulator contamination, and the defect degree includes mild, moderate, and severe. The above is just one example, and there may be other situations, which are not limited here.

[0025] In this embodiment, after step six, the method further includes: When a particle exceeds the boundary after being updated, the particle's position will bounce back to within the boundary. Alternatively, when a particle exceeds the boundary, the portion of the particle that exceeds the boundary will be restored to the boundary. The boundary is predefined.

[0026] For example, the out-of-bounds processing is shown in Equation (10), which uses the method of pulling out-of-bounds particles back to the position of the current optimal particle. The out-of-bounds processing is shown in Equation (10), which takes the cosine position of the solution space as an example.

[0027] , ; in, D This represents the Euclidean distance between the currently globally optimal particle and its nearest neighbor in the solution space. dim It is a particle dimension.

[0028] In this embodiment, after step S4, the method further includes: Based on the output fault labels, the severity of the air switch cabinet fault is determined, and an early warning signal is generated based on the severity.

[0029] In summary, by using multi-parameter monitoring data, including various data related to discharge and defects, the status of the air switchgear can be analyzed from multiple dimensions. By using an improved QPSO (Quantum Particle Swarm Optimization) algorithm to optimize the penalty parameters and kernel function parameters of the SVM (Support Vector Machine) model, the accuracy of fault detection can be greatly improved, effectively avoiding the local optimum problem in traditional algorithms and finding the global optimum. This ensures that the SVM model can accurately classify discharge type, discharge degree, defect type, and defect degree. In conclusion, this application can effectively improve detection accuracy, reduce fault risk, and optimize maintenance strategies, thereby enhancing the operational reliability and safety of the air switchgear, reducing economic losses caused by equipment failure, and improving the stability and efficiency of the entire power system.

[0030] The above combination Figure 1 This application describes the discharge detection method for air switchgear based on QPSO-SVM provided in its embodiments. The following is in conjunction with... Figure 2 This describes the discharge detection device for air switchgear based on QPSO-SVM provided in the embodiments of this application.

[0031] The device specifically includes a data acquisition module and a host computer, as shown below.

[0032] The data acquisition module is used to acquire multi-parameter monitoring data and corresponding fault labels of multiple air switch cabinets to form a training sample set, and to acquire multi-parameter monitoring data of the air switch cabinet under test to form a test dataset. The host computer is used to construct the QPSO-SVM model and initialize the particle swarm of the QPSO algorithm. The particle swarm consists of N particles, each with a position encoded using qubits. Each particle's qubit encoding represents a set of candidate values ​​for the SVM's penalty parameters and kernel function parameters, where N is a positive integer greater than zero. It is also used to input the multi-parameter monitoring data of each sample in the training sample set into the QPSO-SVM model for training, obtaining an optimized QPSO-SVM model. Specifically, training the QPSO-SVM model involves performing the following steps on each particle in the particle swarm: Step 1, based on the qubit encoding corresponding to the i-th particle, decode the specific SVM's penalty parameters and kernel function parameters, configure the SVM, and obtain the configured SVM, where i is a positive integer greater than zero and less than or equal to N; Step 2, input the multi-parameter monitoring data of each sample in the training sample set into the QPSO-SVM model for training. Step 1: Input the configured SVM and map the multi-parameter monitoring data to a high-dimensional feature space using a kernel function. Step 2: Solve for the optimal classification hyperplane in the high-dimensional feature space to construct a classification model. Step 3: Calculate the fitness value of the i-th particle based on the classification accuracy of the classification model for each sample in the training samples. The fitness value is used to characterize the classification accuracy of the configured SVM for the fault labels corresponding to the multi-parameter monitoring data. Step 4: Repeat steps 1 to 4 to obtain the fitness value of each particle. Step 5: Update the individual historical best position and the global best position of the particle swarm based on the fitness value of each particle. Step 6: Adjust the penalty parameters and kernel function of the SVM based on the individual historical best position and the global best position of the particle swarm. Reconfigure the SVM model and repeat steps 1 to 6 until the preset maximum number of iterations is reached to obtain the optimized SVM model. The host computer is also used to input the multi-parameter monitoring data of the air switch cabinet under test into the optimized SVM model for prediction and output the corresponding fault labels.

[0033] In this embodiment, the data measurement module includes an ozone sensor, a carbon monoxide sensor, a nitrogen dioxide sensor, an ultraviolet pulse sensor, and a temperature and humidity smoke sensor. The ozone sensor is used to collect ozone data, the carbon monoxide sensor is used to collect carbon monoxide data, the nitrogen dioxide sensor is used to collect nitrogen dioxide data, the ultraviolet pulse sensor is used to collect ultraviolet pulse data, and the temperature and humidity smoke sensor is used to collect temperature, humidity, and smoke particle concentration data. It is also used to transmit ozone data, carbon monoxide data, nitrogen dioxide data, ultraviolet pulse data, temperature and humidity data, and smoke particle concentration data to the data acquisition module.

[0034] In addition, it should be noted that after the data measurement module transmits ozone data, carbon monoxide data, nitrogen dioxide data, ultraviolet pulse data, temperature and humidity data, and smoke particle concentration data to the data acquisition module, the data acquisition module first performs analog-to-digital conversion in the A / D conversion module to convert it into a format that is easy for the host computer to process.

[0035] In this embodiment, the ozone sensor, carbon monoxide sensor, and nitrogen dioxide sensor are connected to the detachable gas sampling solenoid valve of the air switch cabinet.

[0036] In addition, the device also includes a power module, which includes a battery module and / or a USB charging module.

[0037] It should also be noted that the data acquisition module and the host computer can directly transmit data. The data acquisition module can store data, or a data storage module can be set up between the data acquisition module and the host computer to store the data acquired by the data acquisition module and send it to the host computer.

[0038] In this embodiment, the host computer also includes an early warning software platform, which is used to determine the severity of the air switch cabinet fault based on the output fault label, and generate an early warning signal based on the severity.

[0039] In this embodiment, a Bluetooth communication module and a local early warning module are also included; The local early warning module connects to the host computer via Bluetooth communication to receive early warning signals generated by the host computer and issue alarms via LED lights or buzzers.

[0040] For example, the working process of this device is as follows: 1. Connect the power supply and perform a pre-test. If this is the first time using the device or if it has not been used for a long time, ensure that each sensor warms up for at least 1 hour. If the device has not been used for a short period of time or has been powered off and then restarted, wait for about 10 minutes to ensure that the sensors are working in the best condition. If the battery is low, charge it for backup.

[0041] 2. Turn on the power switch, and the temperature, humidity, and smoke sensors will monitor the data online in real time.

[0042] 3. When the temperature, humidity, and smoke particle concentration data collected by the temperature, humidity, and smoke sensors inside the air switch cabinet are abnormal, the gas sampling solenoid valve will activate, drawing in a gas sample from inside the air switch cabinet. The gas sensor will then operate and obtain gas detection data. Simultaneously, the ultraviolet pulse sensor will operate and obtain corresponding detection data.

[0043] 4. The detected temperature data, humidity data, smoke particle concentration data, ozone and ozone products (carbon monoxide, nitrogen dioxide) gas concentrations and production rates, as well as ultraviolet pulse detection data are stored locally and remotely, and transmitted to the host computer. The switch cabinet status is evaluated using the QPSO-SVM model, and the evaluation results are transmitted to the remote terminal monitoring machine and the local operation and maintenance monitoring machine via Bluetooth communication module and local early warning module, respectively, to provide reference for operation and maintenance personnel.

[0044] Furthermore, the specific implementation of the above-described device is basically similar to the method implementation, so the description is relatively simple. For relevant details, please refer to the description of the method implementation. Moreover, it should be noted that in each module of the device of this application, the components are logically divided according to their intended functions. However, this application is not limited to this and can re-divide or combine the components as needed.

[0045] In another aspect, the present invention provides a computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the methods provided in various optional embodiments of the QPSO-SVM-based air switchgear discharge detection method described above. The computer program is configured to execute the steps in any of the above method embodiments during runtime.

[0046] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims may be performed in a different order than those shown in the embodiments and still achieve the desired result. Furthermore, the specific order or sequential order shown in the drawings is not necessarily required to achieve the desired result; in some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0047] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A discharge detection method for air switchgear based on QPSO-SVM, characterized in that, The method includes: S1. Acquire multi-parameter monitoring data and corresponding fault labels of multiple air switch cabinets to form a training sample set, and collect multi-parameter monitoring data of the air switch cabinet under test to form a test dataset. S2, construct the QPSO-SVM model and initialize the particle swarm of the QPSO algorithm, wherein the particle swarm includes N particles, the position of each particle is encoded by qubits, and the qubit code corresponding to each particle represents a set of candidate values ​​of the penalty parameter and kernel function parameter of the SVM, and N is a positive integer greater than zero; S3, input the multi-parameter monitoring data of each sample in the training sample set into the QPSO-SVM model for training, to obtain the optimized QPSO-SVM model, specifically including: Training the QPSO-SVM model involves performing the following steps for each particle in the particle swarm: Step 1: Based on the qubit encoding corresponding to the i-th particle, decode to obtain the specific SVM penalty parameters and kernel function parameters, configure the SVM, and obtain the configured SVM, where i is a positive integer greater than zero and less than or equal to N; Step 2: Input the multi-parameter monitoring data of each sample in the training samples into the configured SVM, and map the multi-parameter monitoring data to a high-dimensional feature space through the kernel function; Step 3: Solve for the optimal classification hyperplane in the high-dimensional feature space to construct a classification model; Step 4: Based on the classification accuracy of the classification model for each sample in the training samples, calculate the fitness value of the i-th particle, wherein the fitness value is used to characterize the classification accuracy of the configured SVM for the fault labels corresponding to the multi-parameter monitoring data. Step 5: Repeat steps 1 to 4 to obtain the fitness value for each particle; Step 6: Based on the fitness value of each particle, update the individual historical best position of each particle and the global best position of the particle swarm. Step 7: Based on the individual historical best position of each particle and the global best position of the particle swarm, adjust the penalty parameters and kernel function of the SVM, reconfigure the SVM model, and repeat steps 1 to 6 until the preset maximum number of iterations is reached to obtain the optimized SVM model. S4. Input the multi-parameter monitoring data of the air switch cabinet under test into the optimized SVM model for prediction and output the corresponding fault label.

2. The discharge detection method for air switchgear based on QPSO-SVM according to claim 1, characterized in that, The multi-parameter monitoring data includes data on ozone, carbon monoxide, nitrogen dioxide, ultraviolet pulse, temperature and humidity, and smoke particle concentration. The fault label includes at least the discharge type, discharge degree, defect type, and defect degree.

3. The discharge detection method for air switchgear based on QPSO-SVM according to claim 1, characterized in that, Following step six, the following is also included: When a particle exceeds the boundary after being updated, the particle's position will bounce back to within the boundary. Alternatively, when a particle exceeds the boundary, the portion of the particle that exceeds the boundary will be restored to the boundary, where the boundary is pre-defined.

4. The discharge detection method for air switchgear based on QPSO-SVM according to claim 1, characterized in that, After step S4, the method further includes: Based on the output fault labels, the severity of the air switch cabinet fault is determined, and an early warning signal is generated based on the severity.

5. A discharge detection device for air switchgear based on QPSO-SVM, characterized in that, The device includes: The data acquisition module is used to acquire multi-parameter monitoring data and corresponding fault labels of multiple air switch cabinets to form a training sample set, and to acquire multi-parameter monitoring data of the air switch cabinet under test to form a test dataset. The host computer is used to construct a QPSO-SVM model and initialize the particle swarm of the QPSO algorithm. The particle swarm consists of N particles, each with a position encoded using qubits. The qubit encoding for each particle represents a set of candidate values ​​for the SVM's penalty parameters and kernel function parameters, where N is a positive integer greater than zero. The host computer is also used to input the multi-parameter monitoring data of each sample in the training sample set into the QPSO-SVM model for training, obtaining an optimized QPSO-SVM model. Specifically, training the QPSO-SVM model involves performing the following steps on each particle in the particle swarm: Step 1, based on the qubit encoding corresponding to the i-th particle, decode the specific SVM's penalty parameters and kernel function parameters, configure the SVM, and obtain the configured SVM, where i is a positive integer greater than zero and less than or equal to N; Step 2, input the multi-parameter monitoring data of each sample in the training sample set into the host computer. The configured SVM maps multi-parameter monitoring data to a high-dimensional feature space using a kernel function. Step three involves solving for the optimal classification hyperplane in the high-dimensional feature space to construct a classification model. Step four involves calculating the fitness value of the i-th particle based on the classification accuracy of the classification model for each sample in the training samples. The fitness value characterizes the classification accuracy of the configured SVM for the fault labels corresponding to the multi-parameter monitoring data. Step five involves repeating steps one through four to obtain the fitness value of each particle. Step six involves updating the individual historical best position of each particle and the global best position of the particle swarm based on the fitness value of each particle. Step seven involves adjusting the penalty parameters and kernel function of the SVM based on the individual historical best position of each particle and the global best position of the particle swarm, then reconfiguring the SVM model and repeating steps one through six until the preset maximum number of iterations is reached to obtain the optimized SVM model. The host computer is also used to input the multi-parameter monitoring data of the air switch cabinet under test into the optimized SVM model for prediction and output the corresponding fault label.

6. The discharge detection device for air switchgear based on QPSO-SVM according to claim 5, characterized in that, The device includes: The data measurement module includes an ozone sensor, a carbon monoxide sensor, a nitrogen dioxide sensor, an ultraviolet pulse sensor, and a temperature, humidity, and smoke sensor. The ozone sensor is used to collect ozone data, the carbon monoxide sensor is used to collect carbon monoxide data, the nitrogen dioxide sensor is used to collect nitrogen dioxide data, the ultraviolet pulse sensor is used to collect ultraviolet pulse data, and the temperature, humidity, and smoke particle concentration sensor is used to collect temperature, humidity, and smoke particle concentration data. The module also transmits the ozone data, carbon monoxide data, nitrogen dioxide data, ultraviolet pulse data, temperature, humidity, and smoke particle concentration data to the data acquisition module.

7. The discharge detection device for air switchgear based on QPSO-SVM according to claim 6, characterized in that, The ozone sensor, the carbon monoxide sensor, and the nitrogen dioxide sensor are connected to the detachable gas sampling solenoid valve of the air switch cabinet.

8. The discharge detection device for air switchgear based on QPSO-SVM according to claim 5, characterized in that, The host computer is also used to determine the severity of the air switch cabinet fault based on the output fault tags, and to generate an early warning signal based on the severity.

9. The discharge detection device for air switchgear based on QPSO-SVM according to claim 8, characterized in that, The device also includes a Bluetooth communication module and a local early warning module; The local early warning module is connected to the host computer via the Bluetooth communication module, and is used to receive the early warning signal generated by the host computer and to issue an alarm via an LED light or a buzzer.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a computer program or instructions that, when executed, cause the method as described in any one of claims 1-4 to be performed.

Images