Svm-based multi-parameter fusion battery state diagnosis method and system

Abstract

The application discloses a multi-parameter fusion switch battery state diagnosis method and system based on SVM, relates to the field of intelligent monitoring technology combining power line automation and Internet of Things technology, and comprises the following steps: collecting multi-parameter data through a sensor, and pre-processing the multi-parameter data; extracting fault diagnosis features from the processed data, screening important features by using a Pearson correlation coefficient, and dividing the data set into a training set and a test set; combining SVM and a BP neural network to form an SVM-BP model, and evaluating the state of the battery by training and optimizing weights and parameters. The application realizes intelligent operation and maintenance of the battery, changes the battery maintenance work from passive to active, improves the operation and maintenance level of the power distribution automation terminal equipment, greatly reduces the operation and maintenance workload, realizes the state of the backup power battery into the daily operation and management of the power distribution automation equipment, and achieves more timely and rapid operation and maintenance management.

Description

TECHNICAL FIELD

[0001] The application relates to the field of intelligent monitoring technology combining power line automation and Internet of Things technology, and particularly relates to a multi-parameter fusion switch battery state diagnosis method and system based on SVM. BACKGROUND

[0002] With the continuous expansion and in-depth development of the construction of power distribution automation systems, a large number of automation complete switch devices are applied to the production and operation of distribution networks, and continue to grow rapidly with the continuous construction of distribution networks. The backup power supply technology solution of the automation complete switch device is also an important technical problem that must be solved in the construction of power distribution automation. The reasonable configuration of the backup power supply is not only related to the safe operation and power supply reliability of the distribution network, but also affects the construction cost of the power distribution automation and the operation and maintenance cost of the power distribution automation system. This paper mainly analyzes and summarizes the technical requirements, operation and maintenance status and existing problems of the backup battery of the automation complete switch. It is proposed to improve the operation and maintenance management level of the backup battery of the automation complete switch from the aspects of management measures, technical means and supervision management, and support the improvement of the operation and maintenance quality and efficiency of the backup battery of the automation complete switch.

[0003] The backup power supply system of the automation complete switch maintains the continuous work of the distribution terminal for a period of time in the case of line failure or maintenance power failure, to complete a series of work such as fault detection, information reporting and switching on and off operation of the switch. To realize switch state monitoring, analog quantity acquisition, fault judgment, fault isolation and power restoration of non-fault area. Therefore, the backup power supply technology solution of the automation complete switch device is also an important technical problem that must be solved in the construction of power distribution automation. The reasonable configuration of the backup power supply is not only related to the safe operation and power supply reliability of the distribution network, but also affects the construction cost of the power distribution automation and the operation and maintenance cost of the power distribution automation system. At present, the maintenance-free valve-controlled sealed lead-acid battery used in large quantities is not completely maintenance-free, and only correct use and maintenance can have a longer service life. Especially in the harsh environment of intelligent devices in the distribution network, the complex field influencing factors, excessive discharge cycles, excessive discharge depth and the high temperature environment in the southern region in summer all bring about the deterioration of the operation of the lead-acid battery and the reduction of the service life. From the current operation and maintenance experience, there are mainly the following three problems:

[0004] (1) Battery life issues caused by harsh environments. Currently, users require the backup power supply of intelligent power distribution terminals to have as long a lifespan as possible, ideally 8-10 years, especially maintaining a relatively long lifespan even in outdoor ambient temperatures. However, the nominal lifespan of valve-regulated lead-acid batteries is generally around 6 years at 25℃, but in actual operation, performance degradation often occurs in less than 3 years. (2) Battery life issues caused by improper operation and maintenance. Through the summary and analysis of a large number of power distribution automation terminal equipment put into operation on site, a high proportion of the backup power supply systems of current power distribution automation terminals suffer from battery damage or a decrease in actual battery capacity due to improper operation and maintenance such as overcharging and over-discharging. This results in poor battery operation, failure to perform the backup power supply function, and potential hazards to the normal operation of the system. (3) Capacity issues caused by environmental impact. The capacity of valve-regulated lead-acid batteries at -40℃ is only about 30% of that at room temperature, resulting in a sharp decrease in utilization. To address this situation, a derating design approach should be adopted in the engineering design, using more than three times the redundancy capacity in the lead-acid battery capacity configuration to ensure the normal service life of lead-acid batteries under extreme high-temperature climates. Summary of the Invention

[0005] In view of the aforementioned existing problems, this invention employs active battery monitoring technology to implement precise battery maintenance, achieving excellent results. By improving and refining battery monitoring technology, precise monitoring of the battery is implemented, including continuous monitoring of real-time operating data and performance status data. This allows for the verification and measurement of battery capacity and internal resistance. Based on the actual battery condition, automatic maintenance strategies such as charging, discharging, and activation are adjusted to avoid overcharging, over-discharging, and undercharging, achieving intelligent repair and optimized use of the battery, and ensuring precise maintenance of the battery's range. Through real-time monitoring of battery status, verification and measurement of battery capacity and internal resistance, and adjustment of automatic maintenance strategies such as charging, discharging, and activation based on the actual battery condition, intelligent operation and maintenance of the battery can be better realized.

[0006] To address the aforementioned technical issues, a multi-parameter fusion switching battery state diagnosis method based on SVM is proposed, including:

[0007] Multi-parameter data is collected by sensors and preprocessed. Fault diagnosis features are extracted from the processed data, and important features are screened using Pearson correlation coefficient. The dataset is divided into training and test sets. SVM and BP neural network are combined to form SVM-BP model. The weights and parameters are optimized through training to evaluate the battery status.

[0008] As a preferred scheme of the SVM-based multi-parameter fusion switch battery state diagnosis method, the sensor collects multi-parameter data, including multi-parameter data of the equipment under normal operation and various fault states.

[0009] The multi-parameter data includes environmental temperature, environmental humidity, battery surface temperature, working current, working voltage, and battery internal resistance.

[0010] The temperature of the surface of the automatic switch battery and the operating environment temperature of the automatic switch battery are detected by a temperature sensor.

[0011] The humidity of the operating environment is monitored by a humidity sensor.

[0012] The real-time current data of the automatic switch battery is detected by a battery sensor.

[0013] As a preferred scheme of the SVM-based multi-parameter fusion switch battery state diagnosis method, the preprocessing includes cleaning and standardizing the collected data.

[0014] The continuity of the data in time is determined in the time dimension, and when the collected data is continuous in time, the data is preliminarily identified as valid data. The average value of the data in the current time range is calculated, and according to the value range of the data set and the deviation of the average value, the data obviously deviating from the average value, i.e. the identified invalid data and abnormal data, is removed.

[0015] At the same time, when the collected data is discontinuous in the time dimension, the data is identified as random data generated by the collection problem, the values determined as random data are removed, and fixed value filling or statistical value filling is used, and further numerical standardization and normalization processing is performed.

[0016] As a preferred scheme of the SVM-based multi-parameter fusion switch battery state diagnosis method, the preprocessing further includes redundant parameters and complementary parameters obtained by the system through the monitoring device.

[0017] The redundant parameters are multiple repeated information collected by different dimensions of various sensors on any multi-parameter data of the battery joint, and the physical quantity information is collected by various sensors.

[0018] The complementary parameters establish the correlation matrix of the battery temperature and the battery current, the correlation matrix of the battery ambient temperature and the battery internal resistance, the correlation matrix of the battery ambient humidity and the battery internal resistance, calculate the confidence coefficient of the data through the correlation matrix data, merge the confidence coefficient as the complementary parameters of the acquisition parameters into the data set, divide the data set into a training set and a test set, the training set is used for training the model, and the test set is used for evaluating the final effect of the model.

[0019] As a preferred scheme of the SVM-based multi-parameter fusion switch battery state diagnosis method, the method comprises the following steps: preprocessing the data of the switch battery, extracting the important feature set X' from the preprocessed data through the Pearson correlation coefficient, and constructing an automatic switch battery defect diagnosis model by combining the screened important features with historical experience. i The screened features and historical weights are fused to form a comprehensive feature vector F, and an automatic switch battery defect diagnosis model is constructed.

[0020]

[0021] S(F') = Classification (SVM new (F'))

[0022] Wherein, SVM new (F) is a new support vector machine model, b is a bias term, C is a penalty parameter, M is the total number of training samples, ξ j is a relaxation variable, p is a hyperparameter based on historical experience adjustment, S(F') is a state score function, indicating the classification result of the feature vector F'; F' is a comprehensive feature vector of the sample to be classified, Classification indicates the classification process, SVM new (F') is the classification of the feature vector F' using the newly constructed SVM model.

[0023] As a preferred scheme of the SVM-based multi-parameter fusion switch battery state diagnosis method, the method comprises the following steps: preprocessing the data of the switch battery, extracting the important feature set X' from the preprocessed data through the Pearson correlation coefficient, and constructing an automatic switch battery defect diagnosis model by combining the screened important features with historical experience.

[0024] The battery internal resistance Pearson correlation coefficient of the adjacent data in the data set is compared.

[0025] When the solved Pearson correlation coefficient is within the Pearson correlation coefficient threshold range, then the battery internal resistance data corresponding to the maximum value of the Pearson correlation coefficient is recorded as the normal value range of the internal resistance parameter of the current battery pack, the current data set is selected as the health state data set, and the corresponding battery is in a normal state;

[0026] When the correlation coefficient is outside the Pearson correlation coefficient threshold range, then the current value is recorded as an alarm value, and the current data set is marked as a suspected failure data set, corresponding to the battery state being in an alarm state, and further judging the working current of the corresponding battery when the data set is outside the Pearson correlation coefficient threshold range, when the working current is lower than a preset threshold, then the current value is recorded as a failure value, and the current data set is marked as a battery failure data set, corresponding to the battery state being in a failure state.

[0027] As a preferred scheme of the SVM-based multi-parameter fusion switch battery state diagnosis method, the SVM-BP model comprises an improved BP neural network for constructing a weight parameter of an SVM model to form the SVM-BP.

[0028] The BP neural network is trained by a sample, a neural network model of the monitored object is trained, and a model of a logical relationship between an input parameter of the measured object and a running state of the measured object is established.

[0029] The weight and threshold of the BP neural network are determined and optimized: the average value of the square of the difference between the actual monitoring value and the training sample data value is taken as the objective function f(alpha) of the optimization algorithm:

[0030]

[0031] Wherein, P i is the evaluation value of the model output, is the actual value, N is the number of training samples, n is the variable index, and alpha is the normalized value of the BP neural network output result.

[0032] The objective function f(x) of the optimization algorithm of the SVM model is determined:

[0033]

[0034] Wherein, K(x n -x) is a kernel function, and b is a bias term.

[0035] The network is initialized, the input data is provided to the first hidden layer of the network, the data is propagated forward in the network, the neurons of each layer calculate the weighted sum of their inputs, generate the output through the activation function, and deliver it to the next layer, and the last layer, i.e. the output layer, generates the prediction result f(alpha). Through multiple training iterations of the current normalized value to update the alpha value, the SVM model gradually approaches the optimal solution.

[0036] The data of the training set is used for training and optimization of the mathematical model, and the test set data is input into the mathematical model for testing of the mathematical model, and the validity of the model is verified according to the output result;

[0037] The BP neural network is trained according to the sampling data of different running state types of normal, alarm and fault, and is transmitted to the SVM model for model fusion optimization, the accuracy of the model is judged through the evaluation of the output result, and the SVM-BP model is iteratively optimized, the running state of the automatic complete switch battery is evaluated, and the probability of misjudgment and omission is reduced;

[0038] The trained SVM-BP model is deployed into a real-time monitoring system, and the running state of the equipment is monitored and diagnosed in real time, and the real-time monitoring and running state diagnosis of the battery intermediate joint are performed.

[0039] Another object of the present application is to provide a SVM-based multi-parameter fusion switch battery state diagnosis system, which aims to accurately monitor and diagnose the running state of the switch battery, including normal, alarm and fault state; through the fusion of multi-parameter data, the accuracy and reliability of fault diagnosis are improved; by using machine learning algorithm, the probability of misjudgment and omission is reduced, and automatic and intelligent state monitoring is realized; data support is provided for maintenance and management of the switch battery, the service life of the battery is prolonged, and the safe and stable operation of the equipment is ensured.

[0040] As a preferred scheme of the SVM-based multi-parameter fusion switch battery state diagnosis system, it is characterized in that it comprises a data acquisition module, a data preprocessing module, a feature extraction module, a model training and optimization module and a state diagnosis module;

[0041] The data acquisition module collects multi-parameter data of the switch battery through various sensors, including environmental temperature, humidity, battery surface temperature, working current, working voltage and battery resistance, and provides raw data to the data preprocessing module;

[0042] The data preprocessing module receives the raw data of the data acquisition module, cleans, standardizes and normalizes the collected data, and transmits the processed data to the feature extraction module;

[0043] The feature extraction module receives the data of the data preprocessing module, filters out the features corresponding to the fault diagnosis from the preprocessed data by using the Pearson correlation coefficient, assigns weights to the features according to historical experience, constructs a comprehensive feature vector and transmits it to the model training and optimization module;

[0044] The model training and optimization module receives the feature vector of the feature extraction module, combines SVM and BP neural network, constructs an SVM-BP model for training and optimization, trains the model through the training set data, evaluates the model effect using the test set data, and transmits the result to the state diagnosis module.

[0045] The state diagnosis module uses the trained SVM-BP model to score the real-time monitoring data, classifies the battery state as normal, warning and failure, and outputs the diagnosis result.

[0046] A computer device comprises a memory and a processor, and the memory stores a computer program, wherein the processor implements the steps of the SVM-based multi-parameter fusion switch battery state diagnosis method when executing the computer program.

[0047] A computer readable storage medium stores a computer program, wherein the computer program is executed by a processor to implement the steps of the SVM-based multi-parameter fusion switch battery state diagnosis method.

[0048] The present application has the following advantages: the present application can better realize the intelligent operation and maintenance of the battery by real-time monitoring of the battery state, approval and measurement of the battery capacity and internal resistance, and adjustment of the automatic maintenance strategies such as charging, discharging and activation according to the actual state of the battery. Meanwhile, with the real-time monitoring of the battery state, the monitoring state information of the battery can be uploaded to the power distribution automation master station through the information point table of the power distribution automation terminal via the communication channel. When the monitoring finds that the state of the backup power battery is abnormal, the power distribution automation master station is alarmed through remote signaling, and the on-site diagnosis and repair of the operation and maintenance personnel is timely notified.

[0049] The present application realizes the passive to active transition of the battery maintenance work, improves the operation and maintenance level of the power distribution automation terminal equipment, greatly reduces the operation and maintenance workload, realizes the inclusion of the state of the backup power battery into the daily operation and management of the power distribution automation equipment, and achieves more timely and rapid operation and maintenance management. BRIEF DESCRIPTION OF DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0051] Figure 1 The overall flowchart of the SVM-based multi-parameter fusion switch battery state diagnosis method provided by an embodiment of the present application is shown in the figure.

[0052] Figure 2 The system scheme module diagram of the SVM-based multi-parameter fusion switch battery state diagnosis system provided by one embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0053] In order to make the above objectives, features and advantages of the present application more apparent, specific embodiments of the present application are described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work should fall within the protection scope of the present application.

[0054] In the following description, a large number of specific details are set forth in order to facilitate a thorough understanding of the present application, but the present application can also be implemented in other ways different from the description, and those skilled in the art can make similar generalizations without departing from the concept of the present application, so the present application is not limited to the specific embodiments disclosed below.

[0055] Secondly, the "one embodiment" or "embodiment" referred to herein means that a specific feature, structure or characteristic can be included in at least one implementation of the present application. The "in one embodiment" appearing in different places in the specification does not mean the same embodiment, nor does it mean that the embodiment is mutually exclusive with other embodiments.

[0056] The present application is described in detail in combination with the schematic diagram, and in the detailed description of the embodiments of the present application, the cross-sectional view of the device structure is locally enlarged without the general proportion for the convenience of description, and the schematic diagram is only an example, which should not limit the protection scope of the present application herein. In addition, the three-dimensional spatial dimensions of length, width and depth should be included in actual manufacturing.

[0057] Meanwhile, in the description of the present application, it should be noted that the orientation or position relationship indicated by the terms "up, down, inside and outside" is based on the orientation or position relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the indicated device or element must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, the terms "first, second or third" are only for the purpose of description, and cannot be understood as indicating or implying relative importance.

[0058] Unless otherwise defined, the terms "mounting, connecting, associating" in the present application should be interpreted broadly, for example: it can be fixed connection, detachable connection or integral connection; it can also be mechanical connection, electrical connection or direct connection, it can also be indirectly connected through intermediate medium, or it can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0059] Embodiment 1, reference Figure 1 As the first embodiment of the present application, the embodiment provides a multi-parameter fusion switch battery state diagnosis method based on SVM, comprising:

[0060] For the uncontrolled state of the automatic switch battery, it is generally replaced after the battery shows failure, insufficient endurance and inability to meet the backup support demand. The operation and maintenance strategy is simple and extensive, which increases the operation and maintenance workload, wastes the replacement investment cost of the battery, causes resource waste, and cannot fundamentally solve the operation and maintenance problem of the automatic switch battery. At the same time, based on the real-time monitoring of the state of the automatic switch battery, the monitoring state information of the battery can be uploaded to the power distribution automation master station through the information point table of the power distribution automation terminal through the communication channel. When the monitoring finds that the backup power battery state is abnormal, the remote signaling is realized to alarm the power distribution automation master station, and the on-site diagnosis and repair of the operation and maintenance personnel are timely notified. The battery maintenance work is changed from passive to active, which improves the operation and maintenance level of the power distribution automation terminal equipment, greatly reduces the operation and maintenance workload, realizes the state of the backup power battery into the daily operation and management of the power distribution automation equipment, and achieves more timely and rapid operation and maintenance management.

[0061] S1: Collecting multi-parameter data through sensors and pre-processing the multi-parameter data.

[0062] Further, multi-parameter data of the equipment under normal operation and various fault states are collected;

[0063] The multi-parameter data includes environmental temperature, environmental humidity, battery surface temperature, working current, working voltage and battery internal resistance.

[0064] The temperature of the surface of the automatic switch battery and the running environment temperature of the automatic switch battery are detected by a temperature sensor.

[0065] The humidity of the running environment is monitored by a humidity sensor.

[0066] The real-time current data of the automatic switch battery is detected by a battery sensor.

[0067] It should be noted that the collected data is cleaned and standardized to improve the data quality.

[0068] Judging the continuity of data in time dimension, when the collected data is continuous in time, it is preliminarily determined that the data is valid data, the average value of the data in the current time range is calculated, and the data obviously deviating from the average value is removed according to the value range of the data set and the deviation of the average value, that is, the analysis and determination of invalid data and abnormal data;

[0069] At the same time, when the collected data is discontinuous in time dimension, it is determined that the data is random data caused by collection problem, the values determined as random data are removed, and a fixed value is filled, that is, -9999 is used to replace the missing value or statistical value filling, that is, the average number of data set filling, and further standardization and normalization processing of the value.

[0070] The system obtains the information redundancy parameters and complementary parameters through the monitoring device;

[0071] The redundancy parameters are a plurality of repeated information collected by a plurality of sensors through different dimensions on any multi-parameter data of the battery joint, and the physical quantity information is collected by a plurality of sensors;

[0072] The complementary parameters establish the correlation matrix of the battery temperature and the current flowing through the battery, the correlation matrix of the battery environment temperature and the battery internal resistance, and the correlation matrix of the battery environment humidity and the battery internal resistance. The confidence coefficient of the data is calculated through the correlation matrix data, the confidence coefficient is used as the complementary parameter of the collection parameter and is merged into the data set, the data set is divided into a training set and a test set, the training set is used to train the model, and the test set is used to evaluate the final effect of the model.

[0073] S2: Extracting fault diagnosis features from the processed data, screening important features by using Pearson correlation coefficient, and dividing the data set into a training set and a test set.

[0074] Further, the important feature set X' which is helpful to fault diagnosis is extracted from the preprocessed data by Pearson correlation coefficient, including time domain features, frequency domain features and time-frequency domain features. The important features screened are combined with historical experience to construct an automatic complete switch battery defect diagnosis model, and the extracted features are assigned weights to form a weighted feature set W i :

[0075]

[0076] Wherein, W i is the weight of the i-th feature, h i is the historical experience weight corresponding to the feature x i , x i is the i-th feature in the feature set, and I is the total number of features in the feature set.

[0077] The selected features and historical weights are fused to form a comprehensive feature vector An automated complete switch battery defect diagnosis model is constructed, and a state score is obtained through the classification boundary of the model:

[0078]

[0079] S(F') = Classification(SVM new (F'))

[0080] Wherein, SVM new (F) is a new support vector machine model, b is a bias term, C is a penalty parameter, M is the total number of training samples, ξ j is a relaxation variable, p is a hyperparameter based on historical experience adjustment, S(F') is a state score function, indicating the classification result of the feature vector F'; F' is the comprehensive feature vector of the sample to be classified, Classification indicates the classification process, SVM new (F') is the classification of feature vector F' using the newly constructed SVM model.

[0081] It should be noted that the calculation result of constructing an automated complete switch battery defect diagnosis model divides the defects of the automated complete switch battery into three states: normal, warning and failure, and the score value range of different states is determined by the result of S(F').

[0082] By comparing the battery internal resistance Pearson correlation coefficient of adjacent data in the data set;

[0083] When the solved Pearson correlation coefficient is within the Pearson correlation coefficient threshold range, the battery internal resistance data corresponding to the maximum and minimum of the Pearson correlation coefficient is recorded as the normal value range of the internal resistance parameter of the current battery pack, and the current data set is selected as the healthy state data set, corresponding to the normal state of the battery;

[0084] When the correlation coefficient is outside the Pearson correlation coefficient threshold range, the current value is recorded as the warning value, and the current data set is marked as a suspected failure data set, corresponding to the warning state of the battery. Further, the data set with the correlation coefficient outside the Pearson correlation coefficient threshold range is judged according to the working current of the corresponding battery. When the working current is lower than the preset threshold, the current value is recorded as the failure value, and the current data set is marked as a battery failure data set, corresponding to the failure state of the battery.

[0085] S3: Combine SVM with BP neural network to form SVM-BP model, and evaluate the state of the battery by training and optimizing the weight and parameter.

[0086] Further, the improved BP neural network is used to construct the weight parameters of the SVM model to form the SVM-BP;

[0087] The BP neural network is trained by the sample, the neural network model of the monitored object is trained, and the model of the logical relationship between the input parameters of the measured object and the running state of the measured object is established;

[0088] The weight and threshold of the BP neural network are determined and optimized: the average value of the square of the difference between the actual monitoring value and the training sample data value is used as the objective function f(α) of the optimization algorithm:

[0089]

[0090] Where, P i is the evaluation value of the model output, is the actual value, N is the number of training samples, n is the variable index, and α is the normalized value of the BP neural network output result;

[0091] The objective function f(x) of the optimization algorithm of the SVM model is determined:

[0092]

[0093] Where, K(x n -x) is the kernel function, and b is the bias term;

[0094] The network is initialized, the input data is provided to the first hidden layer of the network, the data is propagated forward in the network, the neurons of each layer calculate the weighted sum of their inputs, generate the output through the activation function, and pass it to the next layer. The last layer, i.e. the output layer, produces the prediction result f(α), and the update of the current normalized value to the α value is iterated through multiple training iterations, so that the SVM model gradually approaches the optimal solution;

[0095] The data of the training set is used to train and optimize the mathematical model, and the test set data is input into the mathematical model for testing, and the effectiveness of the model is verified according to the output result;

[0096] The BP neural network is trained for the sampling data of different running state types of normal, alarm and fault, and is transmitted to the SVM model for model fusion optimization. The accuracy of the model can be judged by evaluating the output result, and further iteration optimization of the SVM-BP model is carried out to realize more comprehensive evaluation of the running state of the automatic complete switch battery and reduce the probability of misjudgment and omission. Compared with the traditional single SVM parameter fault diagnosis method, the method combining BP neural network multi-parameter fusion has better classification performance and generalization ability, and the process can further improve the accuracy and reliability of fault diagnosis.

[0097] The trained SVM-BP model is deployed into a real-time monitoring system to monitor and diagnose the running state of the equipment in real time, and to monitor and diagnose the running state of the intermediate joint of the storage battery in real time.

[0098] It should be noted that the above examples are only used to illustrate the technical solutions of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the present application, and they should be covered in the scope of the claims of the present application.

[0099] Embodiment 2, the second embodiment of the present application, is different from the first two embodiments in that:

[0100] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the part of the technical solutions that essentially contribute to the prior art or the part of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0101] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered a list of executable instructions for implementing logic functions, and can be specifically embodied in any computer-readable medium for use by an instruction execution system, apparatus or device, such as a computer-based system, a system including a processor or other system that can fetch and execute instructions from the instruction execution system, apparatus or device, or in conjunction with these instructions. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transport programs for use by an instruction execution system, apparatus or device, or in conjunction with these instructions.

[0102] More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection (electronic) having one or more wires, a portable computer diskette (magnetic), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, and a portable compact disc read-only memory (CDROM). Additionally, the computer readable medium can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example via an optical scanner, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, to generate an electronically readable version of the program, which can then be stored in the computer memory.

[0103] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, various steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any of the following technologies, known in the art, or combinations thereof, can be used: discrete logic circuitry having logic gates for implementing logic functions upon data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), and the like.

[0104] Embodiment 3, with reference to Figure 2 For a third embodiment of the present application, the embodiment provides a SVM-based multi-parameter fusion switchable battery state diagnosis system, including a data acquisition module 10, a data preprocessing module 20, a feature extraction module 30, a model training and optimization module 40, and a state diagnosis module 50;

[0105] The data acquisition module 10 collects multi-parameter data of the switchable battery through various sensors, including environmental temperature, humidity, battery surface temperature, working current, working voltage, and battery resistance, and provides raw data to the data preprocessing module 20;

[0106] The data preprocessing module 20 receives the raw data of the data acquisition module, cleans, standardizes and normalizes the collected data, and transmits the processed data to the feature extraction module 30;

[0107] The feature extraction module 30 receives the data of the data preprocessing module 20, filters out the features corresponding to the fault diagnosis from the preprocessed data using the Pearson correlation coefficient, assigns weights to the features according to historical experience, and constructs a comprehensive feature vector to be transmitted to the model training and optimization module;

[0108] The model training and optimization module 40 receives the feature vector of the feature extraction module 30, combines the SVM and BP neural network, constructs an SVM-BP model for training and optimization, trains the model through the training set data, evaluates the model effect using the test set data, and passes the result to the state diagnosis module 50;

[0109] The state diagnosis module 50 uses the trained SVM-BP model to score the real-time monitoring data, classifies the battery state into normal, warning and failure, and outputs the diagnosis result.

[0110] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A multi-parameter fusion switching battery state diagnosis method based on SVM, characterized in that: include, Multi-parameter data is collected through sensors, and the multi-parameter data is preprocessed. Fault diagnosis features are extracted from the processed data, important features are screened using the Pearson correlation coefficient, and the dataset is divided into training and test sets. By combining SVM with BP neural network to form SVM-BP model, the weights and parameters are optimized through training to evaluate the state of the battery. The sensor collects multi-parameter data, including multi-parameter data under normal operation and various fault conditions of the device; The multi-parameter data includes ambient temperature, ambient humidity, battery surface temperature, operating current, operating voltage, and battery internal resistance. Temperature sensors are used to detect the surface temperature of the automated switch battery and the ambient temperature of the automated switch battery's operating environment. Use a humidity sensor to monitor the humidity of the operating environment; Utilize battery sensors to detect real-time current data of the batteries in automated switchgear. The preprocessing includes cleaning and standardizing the collected data; The continuity of data over time is determined by the time dimension. When the collected data is continuous over time, it is initially identified as valid data. The average value of the data within the current time range is calculated. Based on the deviation between the value range and the average value of the dataset, data that deviates significantly from the average value is removed, i.e., invalid data and abnormal data identified by analysis. Meanwhile, when the collected data is discontinuous in time dimension, it is identified as random data caused by collection problems. The values ​​identified as random data are cleared, and fixed values ​​or statistical values ​​are used for filling. Furthermore, the values ​​are standardized and normalized. The preprocessing also includes redundant and complementary parameters of information obtained by the system through the monitoring device; The redundant parameters are multiple repeated information collected from various sensors on any multi-parameter data of the battery connector through different dimensions, and physical quantity information is collected by multiple sensors. The complementary parameters establish correlation matrices between battery temperature and battery current, battery ambient temperature and battery internal resistance, and battery ambient humidity and battery internal resistance. Confidence coefficients are calculated from the correlation matrix data, and these confidence coefficients are used as complementary parameters to merge into the dataset. The dataset is then divided into a training set and a test set. The training set is used to train the model, and the test set is used to evaluate the final performance of the model.

2. The SVM-based multi-parameter fusion switching battery state diagnosis method as described in claim 1, characterized in that: The extraction of fault diagnosis features includes extracting an important feature set X, which is helpful for fault diagnosis, from the preprocessed data using the Pearson correlation coefficient. ' Including time-domain features, frequency-domain features, and time-frequency-domain features, the selected key features are combined with historical experience to construct an automated complete set of switch-mode battery defect diagnosis models. Based on historical experience, weights are assigned to the extracted features to form a weighted feature set W. i The selected features and historical weights are fused to form a comprehensive feature vector F. An automated complete set of switch battery defect diagnosis model is constructed, and the state score is obtained through the classification boundary obtained by the model. S(F')=Classification(SVM new (F')) Among them, SVM new (F) represents the new support vector machine model, b is the bias term, C is the penalty parameter, M is the total number of training samples, and ξ is the bias term. j Let p be a slack variable, S(F') be a hyperparameter adjusted based on historical experience, and S(F') be a state scoring function representing the classification result of feature vector F'. F' is the comprehensive feature vector of the sample to be classified, and Classification represents the classification process. SVM new (F') is used to classify the feature vector F' using the newly constructed SVM model.

3. The SVM-based multi-parameter fusion switching battery state diagnosis method as described in claim 2, characterized in that: The extraction of fault diagnosis features also includes classifying the defects of the automated complete set of switch batteries into three states: normal, alarm, and fault, based on the calculation results of the automated complete set of switch battery defect diagnosis model, and defining the score range of different states through the results of S(F'). By comparing the Pearson correlation coefficient of the battery internal resistance of adjacent data in the dataset; When the Pearson correlation coefficient is within the threshold range of the Pearson correlation coefficient, the battery internal resistance data corresponding to the extreme value of the Pearson correlation coefficient is recorded as the normal range of the internal resistance parameter of the current battery pack. The current dataset is selected as the health status dataset, corresponding to the normal state of the battery. When the correlation coefficient is outside the Pearson correlation coefficient threshold range, the current value is recorded as an alarm value, and the current dataset is marked as a suspected fault dataset, with the corresponding battery status being an alarm status. Further, for datasets with correlation coefficients outside the Pearson correlation coefficient threshold range, the corresponding battery operating current is determined. When the operating current is lower than a preset threshold, the current value is recorded as a fault value, and the current dataset is marked as a battery fault dataset, with the corresponding battery status being a fault status.

4. The SVM-based multi-parameter fusion switching battery state diagnosis method as described in claim 3, characterized in that: The SVM-BP model includes SVM-BP, which is constructed using weight parameters of an improved BP neural network to build an SVM model. By training samples using a backpropagation (BP) neural network, a neural network model of the monitored object is trained, establishing a model of the logical relationship between the input parameters of the monitored object and the operating state of the monitored object. The weights and thresholds of the BP neural network are determined for optimization: the average of the squared differences between the actual detected values ​​and the training sample data values ​​is used as the objective function f(α) of the optimization algorithm. Among them, P i The evaluation value output by the model. α is the actual value, N is the number of training samples, n is the variable index, and α is the normalized value of the output of the BP neural network. Determine the objective function f(x) of the optimization algorithm for the SVM model: Where K(x) n -x) is the kernel function, and b is the bias term; The network is initialized by providing input data to the first hidden layer. The data propagates forward in the network, and each neuron in each layer calculates the weighted sum of its inputs and generates an output through an activation function, which is then passed to the next layer. The last layer, the output layer, produces the prediction result f(α). By iteratively updating the value of α with the current normalized value through multiple training iterations, the SVM model gradually approaches the optimal solution. The mathematical model is trained and optimized using the training set data, and the mathematical model is tested by inputting the test set data into the mathematical model. The effectiveness of the model is verified based on the output results. A BP neural network is trained on the sampled data of different operating states, including normal, alarm, and fault, and then fed into an SVM model for model fusion optimization. The accuracy of the model is judged by evaluating the output results, and the SVM-BP model is iteratively optimized to evaluate the operating status of the automated switch battery and reduce the probability of misjudgment and missed judgment. The trained SVM-BP model is deployed to the real-time monitoring system to monitor the equipment's operating status and diagnose faults in real time, including real-time monitoring and operating status diagnosis of the battery intermediate connector.

5. A system employing the SVM-based multi-parameter fusion switching battery state diagnosis method as described in any one of claims 1 to 4, characterized in that: It includes a data acquisition module, a data preprocessing module, a feature extraction module, a model training and optimization module, and a state diagnosis module; The data acquisition module collects multi-parameter data of the switching battery through various sensors, including ambient temperature, humidity, battery surface temperature, operating current, operating voltage, and battery internal resistance, and provides raw data to the data preprocessing module. The data preprocessing module receives the raw data from the data acquisition module, cleans, standardizes, and normalizes the collected data, and then transmits the processed data to the feature extraction module. The feature extraction module receives data from the data preprocessing module, uses the Pearson correlation coefficient to filter out features corresponding to fault diagnosis from the preprocessed data, assigns weights to the features based on historical experience, and constructs a comprehensive feature vector to be passed to the model training and optimization module. The model training and optimization module receives the feature vector from the feature extraction module, combines SVM and BP neural networks to construct an SVM-BP model for training and optimization, trains the model using training set data, evaluates the model performance using test set data, and passes the results to the state diagnosis module. The status diagnosis module uses a trained SVM-BP model to score the status of real-time monitoring data, classifies the battery status into normal, alarm, and fault, and outputs the diagnosis results.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the SVM-based multi-parameter fusion switching battery status diagnosis method as described in any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the SVM-based multi-parameter fusion switching battery state diagnosis method as described in any one of claims 1 to 4.

Images