CN-122020350-A - Online fault diagnosis method and system for power exchange station based on Internet of things

CN122020350ACN 122020350 ACN122020350 ACN 122020350ACN-122020350-A

Abstract

The invention discloses a method and a system for online fault diagnosis of a power exchange station based on the Internet of things, which relate to the technical field of fault diagnosis and comprise the steps of collecting observation variable data of the power exchange station for preprocessing, and forming a stabilized data sequence, establishing a structural vector autoregressive SVAR model, calculating a simplified VAR residual error item to determine a structural impact vector, and analyzing unit impacts of different observed variables. According to the method, the stabilized data sequence is modeled into the mutual linear dynamic relation among a plurality of variables by using the SVAR model, so that the interaction of different variables in a complex system can be simulated, the direct causal relation can be reflected, the structural impact vector reflecting external independent impact is solved by combining the fitting residual error item and the structural constraint matrix, key variables are provided for subsequent impact analysis and attribution calculation, the structural impact refines the independent influence of a single event or variable in the system on other variables, and more accurate fault attribution is facilitated.

Inventors

Rong Yanhai
WANG XI
PENG SHUANG
XU CHENGCHENG
YU KAIAN

Assignees

国网电动汽车服务湖北有限公司
武汉市充换电技术有限公司

Dates

Publication Date: 20260512
Application Date: 20251022

Claims (10)

1. The online fault diagnosis method for the power exchange station based on the Internet of things is characterized by comprising the following steps of: Collecting observation variable data of a power exchange station for preprocessing, forming a stabilized data sequence, establishing a structure vector autoregressive SVAR model, calculating a simplified VAR residual error item to determine a structure impact vector, analyzing unit impacts of different observation variables, recursively calculating variable responses after the impacts, analyzing accumulated influence values and calculating the relative contribution rate of each variable; Screening dominant variable labels as dominant label vectors, applying Empirical Mode Decomposition (EMD), extracting frequency component energy calculated by each mode function application Wavelet Packet Transformation (WPT), analyzing comprehensive entropy values, calculating comprehensive entropy change values of each window, and screening abnormal points; Calculating the comprehensive entropy change rate to determine abnormal characteristics and constructing a structural characteristic matrix; And constructing a composite dominant model, predicting the fault probability, generating a log table, and carrying out encryption transmission.
2. The online diagnosis method for the fault of the power exchange station based on the Internet of things as set forth in claim 1, wherein the calculating the relative contribution rate of each variable comprises, Adopting first-order differential processing for each time sequence, and reconstructing a stabilized data sequence according to differential results of different observation variables to establish a structural vector autoregressive SVAR model; using the smoothed data sequence and its lag term as input, and determining the lag order p based on the red-pool information criterion AIC, fitting using least squares to determine the simplified VAR residual term ; Residual terms and structure constraint matrix obtained by fitting Solving a structural impact vector; analyzing unit impacts of different observation variables based on an impact response function IRF, and defining an impact response function; Selecting a time t, determining the impact of an observed variable as unit impact, setting a predicted step number h, recursively using an SVAR model, and calculating the variable response after the impact; And recursively calculating predicted values of the observed variables, calculating the response value of each variable according to the impact response function IRF, performing cumulative calculation to obtain cumulative influence values of different variables on the observed variables, and calculating the relative contribution rate of each variable.
3. The online diagnosis method of the power exchange station fault based on the Internet of things as set forth in claim 2, wherein the calculating the comprehensive entropy change value of each window, screening abnormal points comprises, According to the sum of the mean value and the standard deviation of the relative contribution degree of the observed variable as a contribution threshold, judging the corresponding variable as a dominant variable if the relative contribution degree is greater than or equal to the contribution threshold; the dominant variable at each moment is marked in stages through an attribution method, and a dominant label vector is formed; Determining a corresponding stabilized data sequence according to a dominant tag vector determined by the length of a prediction interval of response, decomposing an EMD (empirical mode decomposition) layer by using an EMD algorithm until the residual amplitude change is no longer obvious, extracting a frequency coefficient obtained by applying wavelet packet transformation WPT to each mode function, and calculating frequency component energy corresponding to each layer of wavelet coefficient; and calculating the comprehensive entropy change value of each window by using a breakpoint detection algorithm according to the frequency component energy, taking the sum of the average value and the double standard deviation of the comprehensive entropy change value in the normal window as an abnormal threshold value, and recording the calculated comprehensive entropy change value as an abnormal point if the calculated comprehensive entropy change value exceeds the abnormal threshold value.
4. The online diagnosis method of power exchange station fault based on the Internet of things as set forth in claim 3, wherein the calculating the comprehensive entropy change rate determines abnormal characteristics and constructing a structured feature matrix comprises, Determining a time window and a comprehensive entropy change value based on the identified abnormal points, and calculating a comprehensive entropy change rate; Determining frequency spectrum energy distribution and normalized relative contribution degree based on the identified abnormal points, and determining energy duty ratio of each frequency band and energy concentrated frequency band; And taking the comprehensive entropy change rate, the frequency band energy duty ratio and the energy concentrated frequency band as abnormal characteristics, and synthesizing the characteristics of all time windows and abnormal points into a structural characteristic matrix.
5. The online diagnosis method of the battery exchange station fault based on the Internet of things according to claim 4, wherein the performing of the fault probability prediction comprises, Constructing a composite dominant model based on a multi-input fusion convolutional neural network, wherein the composite dominant model comprises an input layer, a trunk layer, a fusion layer and an output layer; The input layer inputs according to the structured feature matrix corresponding to the dominant tag vector, the main layer carries out convolution extraction of features, a convolution network is independently constructed for the feature matrix of each dominant tag, and features on a time window are extracted; the fusion layer splices and fuses the characteristics of different leading labels, and outputs fault probability through the output layer; Model training is carried out through training set data, gradient descent optimization is carried out by using an Adam optimizer, parameters of the model are updated, loss of the model is not obviously changed in a continuous iteration process, and iteration is stopped; and predicting the fault probability of the training model according to the structured feature matrix corresponding to the input dominant label vector.
6. The online diagnosis method of the battery exchange station fault based on the Internet of things of claim 5, wherein the collecting battery exchange station observation variable data for preprocessing comprises, Collecting observation variables of a power exchange station, including battery pack end current, battery pack end voltage, ambient temperature, charging and power exchange times, equipment failure times in the station and incoming station electric bicycle numbers; And carrying out unified time stamp and data preprocessing on the observation variable to form a time sequence with uniform sampling, and constructing a data matrix.
7. The online diagnosis method of a power exchange station fault based on the Internet of things according to claim 6, wherein the generating of the log table for encrypted transmission comprises, Generating a log table through a table output tool according to the structured feature matrix corresponding to the dominant label vector and the predicted fault probability, and encrypting data by adopting AES-256; and carrying out safe transmission on the log form by adopting the MQTT, and transmitting the log form to a cloud for log data storage by a wireless transmission technology.
8. An online diagnosis system for a fault of a power exchange station based on the Internet of things, which is based on the online diagnosis method for the fault of the power exchange station based on the Internet of things of any one of claims 1 to 7, is characterized by comprising, The data processing module is used for collecting data of the observation variables of the power exchange station and preprocessing the data; The structure vector autoregressive module is used for establishing an SVAR model, simplifying VAR residual calculation, determining a structure impact vector, analyzing unit impact response and calculating accumulated response analysis and contribution rate; the tag generation module screens dominant variables according to the contribution rate result and generates dominant tag vectors; the signal decomposition module is used for decomposing the time sequence of each dominant tag, performing spectrum decomposition, and performing spectrum feature extraction and entropy analysis; the abnormal point screening module is used for carrying out entropy change analysis and judging abnormal points and constructing a structural feature matrix; The fault prediction module is used for constructing a composite dominant attribute model to predict the fault probability; and the encryption transmission module records the fault type prediction result and the corresponding time window and abnormal spectrum characteristics, encrypts the log form and transmits the log form to the cloud through a secure communication protocol.
9. The computer equipment comprises a memory and a processor, wherein the memory stores a computer program, and the computer equipment is characterized in that the processor realizes the steps of the online fault diagnosis method for the power exchange station based on the Internet of things according to any one of claims 1-7 when executing the computer program.
10. A computer readable storage medium is provided with a computer program, and is characterized in that the computer program is executed by a processor to realize the steps of the online fault diagnosis method for the power exchange station based on the Internet of things as set forth in any one of claims 1 to 7.

Description

Online fault diagnosis method and system for power exchange station based on Internet of things Technical Field The invention relates to the technical field of fault diagnosis, in particular to a method and a system for online fault diagnosis of a power exchange station based on the Internet of things. Background The rapid development of internet of things (IoT) technology provides an important technical support for intelligent operation of a power exchange station, the internet of things can improve the operation efficiency of the power exchange station and reduce the maintenance cost through real-time monitoring, data transmission, fault diagnosis and prediction, the power exchange station is used as a part of a complex power system, the operation environment, the battery pack performance, the charging and power exchange process and the state of equipment in the station are all affected by various variables, such as current, voltage, temperature, the number of equipment faults, the quantity of incoming electric buses and the like, and the observed variables usually show nonlinear complex dynamic characteristics and are easily interfered by external noise or state fluctuation of equipment; However, the existing fault diagnosis method of the power exchange station generally depends on a single variable analysis or a monitoring system based on simple rules, dynamic interaction among complex variables in a nonlinear coupling system is generally difficult to describe, online diagnosis and prediction cannot be realized for sudden faults or long-term hidden faults, global modeling of the complex dynamic system is lacking, strict causal relations are difficult to be established for historical data of each observed variable, so that system response behaviors under multi-factor impact cannot be effectively analyzed, and secondly, the traditional method is mostly limited to single time scale processing, comprehensive analysis on frequency characteristics and time dynamic characteristics of the variables cannot be performed, frequency band energy distribution and time correlation are ignored, and diagnosis accuracy is reduced. Disclosure of Invention The present invention has been made in view of the above-described problems occurring in the prior art. Therefore, the invention provides a method and a system for online diagnosis of a fault of a power exchange station based on the Internet of things, which solve the problems that the prior method for online diagnosis of the fault of the power exchange station generally depends on a single variable analysis or a monitoring system based on simple rules, dynamic interaction among complex variables in a nonlinear coupling system is difficult to be described, online diagnosis and prediction cannot be realized aiming at sudden faults or long-term hidden faults, global modeling of the complex dynamic system is lacking, strict causal relation is difficult to be established aiming at historical data of each observed variable, so that system response behaviors under multi-factor impact cannot be effectively analyzed, and secondly, the prior method is mostly limited to single time scale processing, frequency characteristics and time dynamic characteristics of the variables cannot be comprehensively analyzed, frequency band energy distribution and time relevance are ignored, and diagnosis accuracy is lowered. In order to solve the technical problems, the invention provides the following technical scheme: in a first aspect, the invention provides an online fault diagnosis method for a power exchange station based on the internet of things, which comprises the following steps: Collecting observation variable data of a power exchange station for preprocessing, forming a stabilized data sequence, establishing a structure vector autoregressive SVAR model, calculating a simplified VAR residual error item to determine a structure impact vector, analyzing unit impacts of different observation variables, recursively calculating variable responses after the impacts, analyzing accumulated influence values and calculating the relative contribution rate of each variable; Screening dominant variable labels as dominant label vectors, applying Empirical Mode Decomposition (EMD), extracting frequency component energy calculated by each mode function application Wavelet Packet Transformation (WPT), analyzing comprehensive entropy values, calculating comprehensive entropy change values of each window, and screening abnormal points; Calculating the comprehensive entropy change rate to determine abnormal characteristics and constructing a structural characteristic matrix; And constructing a composite dominant model, predicting the fault probability, generating a log table, and carrying out encryption transmission. As an optimal scheme of the online fault diagnosis method for the power exchange station based on the Internet of things, the method for online fault diagnosis of the power exchange station