Fault diagnosis and self-healing control method, system, equipment and medium for flexible interconnection device of power distribution network

Abstract

The invention discloses a fault diagnosis and self-healing control method, system, equipment and medium for a flexible interconnection device of a power distribution network, and belongs to the technical field of fault diagnosis and control of the power distribution network, and the method comprises the steps: building a system state vector, collecting operation parameters of the power distribution network, and carrying out the weighted fusion; performing quality inspection on the original data based on the acquired operation parameters, and dynamically adjusting model parameters through digital twin synchronization and model parameter updating; deep learning is carried out to carry out multi-level fault feature extraction, and multi-scale feature fusion is carried out; a fault mode identification module is started, and fault probability is calculated for fault classification; evaluating the severity of the fault according to a fault diagnosis result, selecting an optimal control strategy, and carrying out self-healing control; and fault early warning is started to estimate a future fault occurrence probability, and early warning information of different levels is issued to perform preventive maintenance. According to the invention, full-process automation and intelligentization from data perception and intelligent diagnosis to active control and prospective maintenance are realized.

Description

Technical Field

[0001] This invention relates to the field of power distribution network fault diagnosis and control technology, specifically to a fault diagnosis and self-healing control method, system, equipment and medium for a power distribution network flexible interconnection device. Background Technology

[0002] With the continuous improvement of the intelligence level of power distribution networks, flexible interconnection devices play an important role in improving power supply reliability and power quality. However, existing fault diagnosis technologies are mainly based on traditional protection principles and single sensor information, which are difficult to adapt to the complex operating mechanisms and diverse fault modes of flexible interconnection devices. Traditional methods have limited ability to identify weak fault symptoms, often requiring the fault to develop to a certain extent before it can be detected, thus missing the best time for handling.

[0003] Existing technologies have shortcomings in fault location accuracy, particularly in identifying internal component faults and intermittent faults. While traditional expert systems offer good interpretability, the cost of building and maintaining their knowledge bases is high, making it difficult to adapt to the continuous emergence of new fault modes. Existing methods lack self-learning and adaptive capabilities, and cannot continuously optimize diagnostic strategies based on operational experience.

[0004] In terms of self-healing control, traditional methods often employ preset, fixed strategies, lacking comprehensive consideration of fault types and system states, resulting in limited self-healing effectiveness. Existing technologies do not adequately consider the timing and coordination of the fault recovery process, making them prone to secondary faults or system oscillations. Regarding early warning mechanisms, traditional methods are mostly passive diagnostics, lacking proactive prevention capabilities, making it difficult to achieve early fault detection and preventative maintenance. Summary of the Invention

[0005] In view of the above-mentioned problems, the present invention is proposed.

[0006] Therefore, this invention aims to solve the problems of how to establish a multi-level fault feature extraction model, how to achieve intelligent diagnosis through multi-information fusion, how to design an adaptive self-healing control strategy, and how to construct an active fault early warning system.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network, comprising,

[0008] Establish a system state vector, collect and weightedly fuse power distribution network operating parameters, and initialize fault diagnosis model parameters; perform quality checks on raw data based on collected operating parameters, and dynamically adjust model parameters through digital twin synchronization and model parameter updates; perform deep learning to extract multi-level fault features, quantify the contribution of features to fault diagnosis, and perform multi-scale feature fusion; activate the fault mode recognition module, calculate fault probabilities for fault classification, and evaluate the reliability of diagnostic results; assess fault severity based on fault diagnosis results, select the optimal control strategy, and implement self-healing control; activate fault early warning to predict future fault occurrence probabilities, issue different levels of early warning information for preventative maintenance, and conduct regular performance evaluation and optimization.

[0009] As a preferred embodiment of the fault diagnosis and self-healing control method for a flexible interconnection device in a distribution network according to the present invention, the initialization of fault diagnosis model parameters includes: establishing a system state vector, collecting distribution network operating parameters, and weighting and fusing the collected operating parameters.

[0010] Establish a comprehensive monitoring network for flexible interconnected devices, configure corresponding sensors, initialize fault diagnosis model parameters, configure initial parameters for digital twin models, and establish state transition functions for physical entity models.

[0011] As a preferred embodiment of the fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network according to the present invention, the quality inspection includes: continuously collecting operating parameters of the device's operating status, calculating information entropy, performing quality inspection on the raw data, and extracting time-domain, frequency-domain, and time-frequency-domain features.

[0012] To achieve digital twin synchronization, monitor the synchronization error between the virtual model and the physical entity and adjust the model parameters, establish a data cache, and save historical data for trend analysis.

[0013] As a preferred embodiment of the fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network according to the present invention, the multi-scale feature fusion includes feature extraction through a convolutional neural network and calculation of attention mechanism weights to obtain key features.

[0014] Perform feature importance assessment to quantify the contribution of extracted features to fault diagnosis;

[0015] Implement multi-scale feature fusion to integrate information from different time and frequency scales.

[0016] As a preferred embodiment of the fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network according to the present invention, the fault classification includes: activating the fault mode recognition module, calculating the fault probability, and calculating the posterior probability of each fault type under a given observation based on Bayes' theorem.

[0017] The support vector machine decision function is executed, and fault classification is performed using kernel functions and Lagrange multipliers. The neural network output is calculated, and the probability distribution of fault types is obtained through the softmax function.

[0018] Calculate the diagnostic confidence level, assess the reliability of the diagnostic results, initiate expert system rule reasoning, calculate rule strength, fuzzy reasoning output, knowledge certainty factor and rule weight update respectively, and implement multi-model fusion decision-making.

[0019] The beneficial effects of the preferred technical solutions in the embodiments of the present invention are as follows:

[0020] As a preferred embodiment of the fault diagnosis and self-healing control method for a flexible interconnection device in a distribution network according to the present invention, the self-healing control includes assessing the severity of the fault based on the fault diagnosis results and the weighted scores of all fault indicators.

[0021]

[0022] in, The severity of the fault, The weight of the i-th fault indicator. For the i-th fault indicator value, represents the total number of fault indicators, and i is the variable index;

[0023] Select control action selection The optimal control strategy is determined based on the state value function, where, For optimal control action, The value function for performing action a in state s;

[0024] The self-healing time window is calculated as follows: This includes fault detection time, decision-making time, and execution time, among which, Total self-healing time, For fault detection time, For decision-making time, For execution time;

[0025] The control effect was evaluated, and the power recovery effect before and after self-healing was quantified. The control effect evaluation was as follows:

[0026]

[0027] in, For self-healing effect, Power after self-healing Power during a fault Normal power;

[0028] Implement control actions, including protection device activation, load transfer, and parameter adjustment; monitor the self-healing process in real time; automatically switch to backup schemes when self-healing fails or is ineffective; and record self-healing process data.

[0029] The beneficial effects of the preferred technical solution in the embodiments of the present invention are as follows: by quantifying the severity of the fault, selecting the optimal control strategy and evaluating the self-healing effect, the rapid fault recovery and system resilience of the distribution network are achieved, thereby improving the reliability of power supply.

[0030] As a preferred embodiment of the fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network as described in this invention, the step of initiating fault early warning and estimating the probability of future fault occurrence includes: initiating a fault early warning algorithm; calculating health indicators; comprehensively assessing the health level of the equipment based on the normalized state of various parameters; predicting degradation trends; predicting future state evolution based on an exponential smoothing algorithm; calculating fault probability prediction; estimating the probability of future fault occurrence based on an exponential distribution model; dynamically adjusting the early warning threshold; adaptively adjusting the early warning parameters based on the false alarm rate feedback; issuing different levels of early warning information according to the risk level, including equipment status reminders, maintenance suggestions, and emergency warnings; establishing a preventive maintenance plan; analyzing the root causes of faults and proposing improvement suggestions; updating the fault knowledge base and experience data; and improving the diagnostic model and control strategy.

[0031] Regularly perform performance evaluations, calculate key performance indicators such as diagnostic accuracy, recall, F1 score, and self-healing success rate, analyze the diagnostic effectiveness of different fault types and adaptability to different operating scenarios, identify system performance bottlenecks and improvement opportunities, dynamically adjust algorithm parameters and optimize diagnostic strategies based on performance evaluation results, update deep learning models and expert system rule bases, integrate the latest fault characteristics and handling experience, and establish equipment lifecycle health management records.

[0032] The beneficial effects of the preferred technical solution in the embodiments of the present invention are as follows: through health assessment, degradation trend prediction and dynamic early warning mechanism, preventive maintenance and early intervention of failure risk can be realized, thereby extending equipment life and reducing unexpected downtime.

[0033] Another objective of this invention is to provide a fault diagnosis and self-healing control system for a flexible interconnection device in a power distribution network.

[0034] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a fault diagnosis and self-healing control system for a flexible interconnection device in a power distribution network, comprising: an initialization module, a data acquisition and preprocessing module, a multi-level fault feature extraction module, an intelligent fault diagnosis module, a self-healing control module, and a fault early warning module;

[0035] Initialize the module, establish the system state vector, collect and weight the operating parameters of the distribution network, and initialize the fault diagnosis model parameters;

[0036] The data acquisition and preprocessing module performs quality checks on the raw data based on the acquired operating parameters and dynamically adjusts the model parameters through digital twin synchronization and model parameter updates.

[0037] The multi-level fault feature extraction module uses deep learning to extract multi-level fault features, quantifies the contribution of features to fault diagnosis, and performs multi-scale feature fusion.

[0038] The intelligent fault diagnosis module activates the fault mode recognition module, calculates the fault probability to classify the fault, and evaluates the reliability of the diagnosis results.

[0039] The self-healing control module assesses the severity of the fault based on the fault diagnosis results, selects the optimal control strategy, and performs self-healing control.

[0040] The fault warning module initiates fault warnings to estimate the probability of future faults, issues warning information of different levels for preventive maintenance, and performs regular performance evaluations and optimizations.

[0041] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of the fault diagnosis and self-healing control method of a flexible interconnection device for a power distribution network.

[0042] The present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the steps of the fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network.

[0043] The beneficial effects of this invention are: This invention offers high diagnostic accuracy, employing multi-information fusion and deep learning technologies to achieve a fault diagnosis accuracy rate of 96.8%, a 23.3% improvement compared to traditional methods. The multi-level feature extraction algorithm can identify subtle fault signs, enabling early fault detection and precise fault location.

[0044] The system boasts a rapid response time, with the average diagnosis time reduced to 18 milliseconds and self-healing time kept below 2 seconds, significantly improving its rapid recovery capabilities. The intelligent decision-making mechanism reduces the need for manual intervention and enhances emergency response efficiency.

[0045] With strong self-learning capabilities, the transfer learning technology reduces adaptation time to new scenarios by more than 70%, and the system can continuously optimize diagnostic and control strategies based on operational experience. The digital twin platform provides a secure and controllable verification environment for algorithm training.

[0046] High reliability is ensured by multi-model fusion decision-making and expert system verification mechanisms, reducing the false alarm rate to 2.3%. The early warning mechanism realizes the transformation from passive diagnosis to proactive prevention, providing important technical support for the safe and reliable operation of flexible interconnection devices in the distribution network, and is of great significance for improving the power supply quality and operating efficiency of the distribution network. Attached Figure Description

[0047] To more clearly illustrate the technical solutions of the embodiments of the present invention, 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 the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is a flowchart illustrating the overall process of a fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network, as provided in one embodiment of the present invention. Detailed Implementation

[0049] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0050] Example 1, referring to Figure 1 This is one embodiment of the present invention, which provides a fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network, comprising:

[0051] S100. Establish the system state vector, collect the power distribution network operation parameters and perform weighted fusion, and initialize the fault diagnosis model parameters.

[0052] S200: Based on the collected operating parameters, the raw data is inspected for quality, and the model parameters are dynamically adjusted through digital twin synchronization and model parameter updates.

[0053] S300: Perform deep learning to extract multi-level fault features, quantify the contribution of features to fault diagnosis, and perform multi-scale feature fusion.

[0054] S400: Start the fault mode recognition module, calculate the fault probability, classify the fault, and evaluate the reliability of the diagnostic results.

[0055] S500: Assess the severity of the fault based on the fault diagnosis results, select the optimal control strategy, and perform self-healing control.

[0056] S600 initiates fault early warning to estimate the probability of future faults, issues early warning information of different levels for preventive maintenance, and performs regular performance evaluation and optimization.

[0057] It should be noted that there are technical problems in the fault diagnosis and self-healing control of flexible interconnection devices in power distribution networks, such as low diagnostic accuracy, slow response speed, weak self-healing ability, and imperfect early warning mechanism.

[0058] Therefore, in response to the above-mentioned problems, the steps of S100-S600 have achieved full-process automation and intelligence from data perception and intelligent diagnosis to proactive control and forward-looking maintenance, solving the problems of low diagnostic accuracy, slow response speed, weak self-healing ability, and imperfect early warning mechanism.

[0059] Example 2, refer to Figure 1 This is one embodiment of the present invention, which provides a fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network, comprising:

[0060] In this embodiment of the invention, S100 establishes a system state vector, collects and weights distribution network operating parameters, and initializes fault diagnosis model parameters, including the following steps S101-S102:

[0061] S101. Establish a system state vector, collect power distribution network operating parameters, and perform weighted fusion of the collected operating parameters;

[0062] After the flexible interconnection device for the distribution network is started, a system state vector is established, and voltage, current, temperature, and humidity signals and equipment status parameters of each monitoring point are collected. The specific system state vector is as follows:

[0063]

[0064] in, Let i be the state vector of the i-th monitoring point. It is a voltage signal. It is a current signal. For temperature signals, This is a humidity signal. Here, 'i' is the device status parameter, and 'i' is the variable index, which is a normal number.

[0065] The multi-sensor fusion weights are calculated, and the fusion coefficients are determined based on the measurement uncertainties of each sensor. The specific multi-sensor fusion weights are as follows:

[0066]

[0067] in, Let the fusion weights be those of the j-th sensor. Let represent the measurement uncertainty of the j-th sensor, n be the total number of sensors, and j be the variable index that is a positive constant.

[0068] To achieve state estimation after fusion, weighted fusion of multi-sensor data is performed, resulting in the following state estimation:

[0069]

[0070] in, These are the estimated states after fusion. Let be the measurement value of the j-th sensor.

[0071] S102. Establish a comprehensive monitoring network for flexible interconnected devices, configure corresponding sensors, initialize fault diagnosis model parameters, configure initial parameters for digital twin models, and establish state transition functions for physical entity models.

[0072] Establish a comprehensive monitoring network for flexible interconnected devices, and configure multiple types of sensors such as voltage, current, temperature, and vibration sensors.

[0073] Initialize the fault diagnosis model parameters, including neural network weights, expert system rule base, etc.

[0074] Configure the initial parameters of the digital twin model and establish the state transition function of the physical entity model:

[0075]

[0076] Where x is the state variable and u is the input variable. Here, f represents the model parameters, and f is the state transition function.

[0077] In this embodiment of the invention, step S200 involves quality inspection of the raw data based on the collected operating parameters, and dynamic adjustment of the model parameters through digital twin synchronization and model parameter updates, including the following steps S201-S202:

[0078] In an embodiment of the present invention, S201 involves calculating the operating parameters of the continuous data acquisition device, calculating the information entropy, performing quality checks on the raw data, and extracting time-domain, frequency-domain, and time-frequency-domain features, including the following steps A1-A3:

[0079] A1. Continuously acquire multi-dimensional data on the operating status of the device, assess the information content and uncertainty of the data, and calculate the information entropy:

[0080]

[0081] in, For information entropy, Let be the probability of the i-th state. This represents the total number of states.

[0082] A2. Perform quality checks on the raw data, including outlier detection, noise filtering, and missing value imputation.

[0083] A3. Implement data standardization and feature engineering to extract time-domain, frequency-domain, and time-frequency-domain features.

[0084] In an optional implementation, the quality inspection in S201 can be a quality inspection method based on statistical thresholds and simple filtering. This involves continuously collecting multi-dimensional data on the device's operating status, pre-setting normal range thresholds for each parameter, checking in real time whether each data point exceeds the threshold, marking outliers, and removing or replacing marked outliers with the average of adjacent data. Moving average filtering is applied to smooth noisy data. For missing values, the previous valid value is used to fill them. Time-domain and frequency-domain features are calculated on the cleaned data, but complex time-frequency domain analysis is not performed; only basic frequency-domain components are retained. However, this method has low sensitivity for detecting dynamic changes or sudden faults, leading to the loss of subtle fault features.

[0085] In another optional implementation, the quality inspection in S201 can also be a quality inspection method based on a rule engine and consistency verification. This method continuously collects multi-dimensional data on the device's operating status, pre-sets a set of simple rules, and applies the rule engine to each data stream based on expert experience or equipment manuals to check whether the data violates the rules. Simultaneously, it compares the consistency of data from multiple sensors, such as verifying whether the phases of voltage and current match. For inconsistent data, it uses majority voting or historical data repair. On the data that passes the verification, it extracts time-domain features (such as the maximum, minimum, and mean values ​​within a sliding window) and frequency-domain features (extracting the dominant frequency through power spectrum analysis), but does not perform information entropy calculation or in-depth time-frequency domain transformation; however, the feature extraction range is limited.

[0086] S202. Perform digital twin synchronization, monitor the synchronization error between the virtual model and the physical entity, adjust the model parameters, establish a data cache, and save historical data for trend analysis.

[0087] Perform digital twin synchronization:

[0088]

[0089] in, For synchronization error, This is a virtual model state. It represents the physical entity's state; it enables digital twin synchronization and monitors the synchronization error between the virtual model and the physical entity's state.

[0090] Model parameter updates are performed by dynamically adjusting the model parameters based on the synchronous loss gradient. The model parameter updates are as follows:

[0091]

[0092] in, For the updated parameters, For learning rate, To synchronize the loss gradient.

[0093] Establish a data caching mechanism to save historical data for trend analysis and ensure data security mechanisms to prevent data tampering and leakage.

[0094] In an embodiment of the present invention, deep learning is performed in S300 to extract multi-level fault features, quantify the contribution of features to fault diagnosis, and perform multi-scale feature fusion, including the following steps S301-S303:

[0095] S301. Feature extraction is performed through a convolutional neural network, and key features are obtained by calculating the weights of the attention mechanism.

[0096] The deep learning feature extraction module is activated to perform convolutional neural network feature extraction, extracting spatial features through convolution operations and activation functions:

[0097]

[0098] in, For the k-th feature map of the l-th layer, The weights are the kernel weights, and * represents the convolution operation. For bias, This is the activation function.

[0099] Performing recursive computation using an LSTM network to capture temporal features and fault evolution patterns: LSTM network recursive computation:

[0100]

[0101] in, Let be the hidden state at time t. For the input weight matrix, This is a cyclic weight matrix. For the input vector, This is the bias vector.

[0102] The attention mechanism weights are calculated to highlight key features and improve the targeting of feature extraction. Specifically, the attention mechanism weights are:

[0103]

[0104] in, For attention weights, The attention score is given by T, where T is the sequence length and k is the variable index.

[0105] S302. Conduct feature importance assessment, including the following steps B1:

[0106] B1. The contribution of quantified extracted features to fault diagnosis:

[0107]

[0108] in, For the importance of the k-th feature, Let k be the feature value of the i-th sample. denoted as the feature mean, and N as the number of samples.

[0109] In an optional implementation, the importance assessment in S302 can be an assessment method based on the frequency of model feature usage. During the training of the fault diagnosis model, a tree-like model such as a decision tree or random forest is used to record the number of times each feature is selected as a split node when constructing the decision tree; the total frequency of each feature used in all trees is counted, and the higher the frequency, the greater the contribution of the feature to fault classification; a list of feature importance is generated according to the frequency, which is used for subsequent fault diagnosis optimization and feature selection; however, if the fault diagnosis model is not a tree-like model (such as a neural network), the assessment results are inaccurate or difficult to apply.

[0110] In another optional implementation, the importance assessment in S302 can also be an assessment method based on the correlation between features and fault labels, which calculates the statistical correlation between each feature and the fault label to measure the strength of the association between the feature and the occurrence of the fault; the features are ranked according to the absolute value of the correlation, and the features with higher correlation are considered to be more important; the ranked feature importance is integrated into the fault diagnosis model for weighting or screening key features; however, it is insufficient for assessing the nonlinear relationships (such as transient faults) that are common in distribution network faults.

[0111] S303. Implement multi-scale feature fusion to integrate information from different time and frequency scales;

[0112] Implement multi-scale feature fusion to integrate information from different time and frequency scales.

[0113] Transfer learning techniques are employed to calculate the similarity between the source and target domains, the transfer loss function, feature domain adaptation, and knowledge distillation. Specifically, the similarity between the source and target domains is as follows:

[0114]

[0115] in, For source domain With the target domain similarity, , Let d be the value of the i-th feature in the source and target domains, and d be the feature dimension.

[0116] The migration loss function is:

[0117]

[0118] in, For migration loss, For mission losses, For domain adaptation loss, These are the balancing parameters.

[0119] The feature domain is adapted as follows:

[0120]

[0121] Wherein, MMD represents the maximum mean difference. For feature mapping function, , This represents the number of samples in the source and target domains.

[0122] The knowledge distillation is as follows:

[0123]

[0124] in, For knowledge distillation loss, For hard label loss, The value represents the soft tag loss, and T is the temperature parameter.

[0125] In this embodiment of the invention, the fault mode recognition module is activated in step S400 to calculate the fault probability, classify the fault, and evaluate the reliability of the diagnostic results, including the following steps S401-S403:

[0126] S401. Start the fault mode recognition module, calculate the fault probability, and calculate the posterior probability of each fault type under given observations based on Bayes' theorem.

[0127] The fault mode identification module is activated to calculate the fault probability. Based on Bayes' theorem, the posterior probability of each fault type under given observations is calculated:

[0128]

[0129] in, For a given observation X, the fault The posterior probability, Let be the likelihood probability. For prior probability, Let k be the total number of faults and k be the variable index.

[0130] S402. Execute the support vector machine decision function, use the kernel function and Lagrange multipliers to classify faults, calculate the neural network output, and obtain the probability distribution of fault types through the softmax function.

[0131] The support vector machine (SVM) decision function is executed, using kernel functions and Lagrange multipliers for fault classification. The specific SVM decision function is as follows:

[0132]

[0133] in, Let be the decision function. For Lagrange multipliers, For class tags, is the kernel function, b is the bias term, and l is the number of support vectors.

[0134] The neural network output is calculated, and the probability distribution of fault types is obtained through the softmax function. The neural network output is:

[0135]

[0136] Where y is the output probability vector. For output layer weights, This is the last hidden layer. This is the output layer bias.

[0137] S403. Calculate the diagnostic confidence level, assess the reliability of the diagnostic results, initiate expert system rule reasoning, calculate the rule strength, fuzzy reasoning output, knowledge certainty factor and rule weight update respectively, and implement multi-model fusion decision-making.

[0138] Calculate the diagnostic confidence level and assess the reliability of the diagnostic results:

[0139]

[0140] Where C represents the diagnostic confidence level. This represents the maximum probability of failure.

[0141] Initiate rule-based reasoning in the expert system, calculating rule strength, fuzzy reasoning output, knowledge certainty factor, and updating rule weights. Specifically, rule strength is calculated as follows:

[0142]

[0143] in, Let r be the activation strength of rule r. Let be the membership function value for the i-th condition. This represents the total number of conditions for rule-based reasoning.

[0144] Fuzzy inference output:

[0145]

[0146] in, Let R be the membership degree of the conclusion, and R be the set of rules. Let r be the membership degree of the conclusion part of rule r.

[0147] Knowledge certainty factor:

[0148]

[0149] Where CF is the deterministic factor. Let P(H) be the probability of hypothesis H given evidence E, and let P(H) be the prior probability of the hypothesis.

[0150] Rule weight update:

[0151]

[0152] in, For the updated weights, The original weights, For learning rate, is the error signal, and x is the input.

[0153] Implement multi-model fusion decision-making, integrate the diagnostic results of different algorithms, and provide diagnostic basis and reasoning process.

[0154] In this embodiment of the invention, step S500 assesses the severity of the fault based on the fault diagnosis results, selects the optimal control strategy, and performs self-healing control, including the following steps S501-S503:

[0155] S501. Based on the fault diagnosis results, assess the severity of the fault by comprehensively calculating the weighted scores of all fault indicators:

[0156]

[0157] in, The severity of the fault, The weight of the i-th fault indicator. For the i-th fault indicator value, represents the total number of fault indicators, and i is the variable index;

[0158] S502, Select Control Action Selection:

[0159]

[0160] The optimal control strategy is determined based on the state value function, where... For optimal control action, The value function for performing action a in state s;

[0161] The self-healing time window is calculated by including fault detection time, decision-making time, and execution time, specifically:

[0162]

[0163] in, Total self-healing time, For fault detection time, For decision-making time, For execution time;

[0164] S503. Conduct a control effect evaluation, quantify the power recovery effect before and after self-healing, and evaluate the control effect as follows:

[0165]

[0166] in, For self-healing effect, Power after self-healing Power during a fault Normal power;

[0167] Implement control actions, including protection device activation, load transfer, and parameter adjustment; monitor the self-healing process in real time; automatically switch to backup schemes when self-healing fails or is ineffective; and record self-healing process data.

[0168] In an embodiment of the present invention, S600 initiates a fault early warning to estimate the probability of future faults, issues early warning information of different levels for preventive maintenance, and performs periodic performance evaluation and optimization, including the following steps S601-S602:

[0169] S601. Activate the fault early warning algorithm, calculate the health index, and comprehensively assess the equipment health level based on the normalized status of all parameters. Health index:

[0170]

[0171] in, As a health indicator, For the i-th parameter value, , Let be the minimum and maximum values ​​of the parameters, and b be the total number of all parameters.

[0172] To predict degradation trends, the future state evolution is predicted based on an exponential smoothing algorithm.

[0173]

[0174] in, The predicted value at time t+k. For smoothing coefficients, For the current observation value, This is the current estimate.

[0175] Calculate failure probability prediction and estimate the probability of future failures based on an exponential distribution model:

[0176]

[0177] in, For the future Failure probability over time This refers to the failure rate.

[0178] The warning threshold is dynamically adjusted.

[0179]

[0180] in, For the new threshold, The original threshold, To adjust the coefficient, For false alarm rate, The target is the false alarm rate.

[0181] In an embodiment of the present invention, issuing early warning information at different levels for preventative maintenance includes the following steps C1-C3:

[0182] C1. Implement dynamic adjustment of early warning thresholds and adaptively adjust early warning parameters based on false alarm rate feedback.

[0183] C2. Issue different levels of early warning information based on the risk level, including equipment status reminders, maintenance suggestions, and emergency warnings.

[0184] C3. Establish a preventive maintenance plan, analyze the root causes of failures and propose improvement suggestions, update the fault knowledge base and experience data, and improve the diagnostic model and control strategy.

[0185] In an optional implementation, the preventive maintenance of S601 can be based on moving averages and linear regression. This involves collecting equipment operating parameters, normalizing each parameter, and then calculating a comprehensive health index using a weighted average method, with weights manually set according to parameter importance. A simple moving average method is used to smooth historical health indices, eliminating random fluctuations. A linear regression model is applied to fit the smoothed data to predict the health trend over a future period. Based on historical fault records, a Weibull distribution model is used to estimate the equipment failure rate, and combined with the current health trend, the probability of future failures is calculated. Based on recent false alarm rate statistics, a rule-based adjustment method is used to dynamically modify the warning threshold; for example, if the false alarm rate is too high, the threshold is appropriately increased, and vice versa. Based on the comparison between the predicted failure probability and the threshold, different levels of warning information are issued, and preventive maintenance recommendations are generated.

[0186] In another optional implementation, the preventive maintenance of S601 can also be based on time series analysis and statistical processes. This involves collecting multi-dimensional operating parameters, using principal component analysis (PCA) to reduce dimensionality and extract key features, constructing a comprehensive health index to reflect the overall equipment status, and employing an autoregressive integral moving average (ARIMA) model to perform time series analysis on the health index, identifying seasonal and trend changes, and predicting future health evolution. Based on equipment reliability theory, a log-normal distribution model combined with historical fault data is used to estimate the probability of failure within a specific future time window. Statistical process control (SPC) methods are applied to set control limits (such as the 3σ principle) as early warning thresholds, and the control limits are dynamically adjusted according to the process capability index to reduce false alarms. When the health index exceeds the control limits or the failure probability exceeds the set value, an early warning system is automatically triggered, issuing tiered early warning information and recommending preventive maintenance measures (such as equipment calibration and optimization of operating parameters).

[0187] In an embodiment of the present invention, S602, performing periodic performance evaluation and optimization, includes the following steps D1-D5:

[0188] D1. Regularly perform performance evaluations, calculating key performance indicators such as diagnostic accuracy, recall, F1 score, and self-healing success rate. Specifically, diagnostic accuracy:

[0189]

[0190] Where Accuracy is the accuracy rate, TP is the true positives, TN is the true negatives, FP is the false positives, and FN is the false negatives.

[0191] Recall rate:

[0192]

[0193] Here, Recall is the recall rate.

[0194] F1 score:

[0195]

[0196] Where F1 is the F1 score. For accuracy.

[0197] Self-healing success rate:

[0198]

[0199] in, To increase the success rate of self-healing, To the number of times the self-healing is successful, This represents the total number of failures.

[0200] D2. Analyze the diagnostic effectiveness of different fault types and their adaptability to different operating scenarios, and identify system performance bottlenecks and areas for improvement.

[0201] D3. Based on the performance evaluation results, dynamically adjust the algorithm parameters and optimize the diagnostic strategy.

[0202] D4. Update the deep learning model and expert system rule base, and integrate the latest fault characteristics and handling experience.

[0203] D5. Establish a health management record for the entire life cycle of equipment, improve the monitoring and early warning mechanism, and provide quantitative basis and decision support for continuous system improvement.

[0204] In one optional implementation, the periodic performance evaluation and optimization of S602 can involve collecting historical operational data on fault diagnosis and self-healing control at fixed intervals, including the number of fault occurrences, the accuracy of diagnostic results, and the success or failure of self-healing actions; calculating simple performance indicators based on the collected data, such as the diagnostic accuracy rate (the ratio of the number of correct diagnoses to the total number of diagnoses) and the self-healing success rate (the ratio of the number of successful self-heals to the total number of faults); manually adjusting the threshold parameters (such as the fault probability threshold) or control strategy parameters (such as the self-healing action triggering conditions) of the fault diagnosis model by operators according to the changing trends of the performance indicators to optimize performance; manually updating the expert system rule base based on historical fault handling experience and operator feedback, adding common fault modes or modifying existing rules; and maintaining an equipment health record table to record each performance evaluation result, adjustment measures, and fault events for subsequent trend analysis and reference; however, this approach cannot adapt to complex fault scenarios and has low optimization efficiency.

[0205] In another optional implementation, the periodic performance evaluation and optimization of S602 can also be performed as follows: a comprehensive performance review is conducted every quarter or semi-annually, collecting operational data and analysis reports of the fault diagnosis model and self-healing control strategy; system status is evaluated using predefined performance evaluation rules (e.g., a diagnostic error rate exceeding 10% or a self-healing success rate below 90% is considered substandard), without involving complex statistical calculations; an expert review meeting is organized to discuss the performance evaluation results, identify problem areas (e.g., poor diagnostic performance for specific fault types), and decide on optimization measures (e.g., adjusting model parameters or changing control strategies); based on the meeting decisions, the training dataset of the deep learning model is manually updated, new fault cases are added, or the expert system rule base is adjusted, but automatic transfer learning or knowledge distillation is not used; an equipment status and performance database is established to record key parameters, fault history, and optimization records for simple trend analysis and preventive maintenance plan development; however, this approach suffers from high response latency and lacks real-time adaptive capabilities.

[0206] Example 3 is an embodiment of the present invention, which provides a fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0207] A simulation platform for flexible interconnected devices, including back-to-back converters, DC support capacitors, and control and protection systems, was established.

[0208] The verification was conducted based on actual operating data from a certain region and manually injected faults, taking into account various influencing factors such as device aging, environmental changes, and load fluctuations.

[0209] Table 1 Performance Comparison of Different Fault Diagnosis Methods

[0210] Diagnostic methods Accuracy (%) Recall rate (%) F1 score Diagnosis time (ms) False alarm rate (%) Traditional preservation 78.5 72.3 0.753 15 8.2 Expert system 85.2 80.6 0.828 45 5.8 Machine Learning 91.3 87.9 0.896 25 4.1 Method of the present invention 96.8 94.5 0.957 18 2.3

[0211] Table 2 Analysis of Diagnostic Effectiveness for Different Fault Types

[0212] Fault type Detection accuracy (%) Positioning accuracy (%) Average detection time (ms) Self-healing success rate (%) Switching device failure 97.2 95.8 12 92.5 Capacitor failure 95.6 93.4 18 87.3 Sensor failure 98.1 96.7 8 95.8 communication failure 94.3 91.6 25 89.2 Control system failure 96.5 94.1 15 91.7

[0213] Table 3 Adaptability Analysis for Different Operating Scenarios

[0214] Operating scenarios Load factor (%) Diagnostic accuracy (%) Self-healing time (s) System recovery rate (%) Light load operation <30 94.8 1.2 96.5 Normal operation 30-80 97.1 0.8 98.2 Heavy load operation >80 95.3 1.5 94.8 Fault recovery change 92.7 2.1 91.3

[0215] Simulation results show that the fault diagnosis and self-healing control method for flexible interconnection devices in power distribution networks proposed in this invention achieves significant improvements in all performance indicators.

[0216] The fault diagnosis accuracy reached 96.8%, the average diagnosis time was shortened to 18 milliseconds, and the false alarm rate was reduced to 2.3%.

[0217] In the analysis of diagnostic effectiveness for different fault types, the diagnostic accuracy for sensor faults was the highest, reaching 98.1%, with a self-healing success rate of 95.8%.

[0218] Adaptability analysis under different operating scenarios shows that the present invention performs best under normal operating conditions, with a system recovery rate of 98.2%.

[0219] The transfer learning mechanism reduces the need for labeled data in new scenarios by more than 70%, fully verifying the effectiveness and practicality of the present invention.

[0220] Example 4 is an embodiment of the present invention. The above is an illustrative scheme of a fault diagnosis and self-healing control method for a flexible interconnection device in a distribution network. It should be noted that the technical solution of a fault diagnosis and self-healing control system for a flexible interconnection device in a distribution network and the technical solution of the fault diagnosis and self-healing control method for a flexible interconnection device in a distribution network described above belong to the same concept. Details not described in detail in the technical solution of the fault diagnosis and self-healing control system for a flexible interconnection device in a distribution network in this embodiment can be found in the description of the technical solution of the fault diagnosis and self-healing control method for a flexible interconnection device in a distribution network described above.

[0221] This embodiment provides a fault diagnosis and self-healing control system for a flexible interconnection device in a power distribution network, including: an initialization module, a data acquisition and preprocessing module, a multi-level fault feature extraction module, an intelligent fault diagnosis module, a self-healing control module, and a fault early warning module.

[0222] Initialize the module, establish the system state vector, collect and weight the operating parameters of the distribution network, and initialize the fault diagnosis model parameters;

[0223] The data acquisition and preprocessing module performs quality checks on the raw data based on the acquired operating parameters and dynamically adjusts the model parameters through digital twin synchronization and model parameter updates.

[0224] The multi-level fault feature extraction module uses deep learning to extract multi-level fault features, quantifies the contribution of features to fault diagnosis, and performs multi-scale feature fusion.

[0225] The intelligent fault diagnosis module activates the fault mode recognition module, calculates the fault probability to classify the fault, and evaluates the reliability of the diagnosis results.

[0226] The self-healing control module assesses the severity of the fault based on the fault diagnosis results, selects the optimal control strategy, and performs self-healing control.

[0227] The fault warning module initiates fault warnings to estimate the probability of future faults, issues warning information of different levels for preventive maintenance, and performs regular performance evaluations and optimizations.

[0228] This embodiment also provides an electronic device applicable to a fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network as proposed in the above embodiment.

[0229] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements a fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network as proposed in the above embodiments.

[0230] The storage medium proposed in this embodiment and the fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0231] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0232] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network, characterized in that: include, Establish a system state vector, collect and weightedly fuse power distribution network operating parameters, and initialize fault diagnosis model parameters; The raw data is quality checked based on the collected operating parameters, and the model parameters are dynamically adjusted through digital twin synchronization and model parameter updates. Deep learning is used to extract multi-level fault features, quantify the contribution of features to fault diagnosis, and perform multi-scale feature fusion. The fault mode recognition module is activated to calculate the fault probability, classify the faults, and assess the reliability of the diagnostic results. The severity of the fault is assessed based on the fault diagnosis results, and the optimal control strategy is selected for self-healing control. Initiate fault warnings to estimate the probability of future faults, issue warning information at different levels for preventive maintenance, and conduct regular performance evaluations and optimizations.

2. The fault diagnosis and self-healing control method for a flexible interconnection device in a distribution network as described in claim 1, characterized in that: The initialization of fault diagnosis model parameters includes establishing a system state vector, collecting power distribution network operating parameters, and weighting and fusing the collected operating parameters. Establish a comprehensive monitoring network for flexible interconnected devices, configure corresponding sensors, initialize fault diagnosis model parameters, configure initial parameters for digital twin models, and establish state transition functions for physical entity models.

3. The fault diagnosis and self-healing control method for a flexible interconnection device in a distribution network as described in claim 2, characterized in that: The quality inspection includes continuously acquiring the operating parameters of the device, calculating the information entropy, performing quality inspection on the raw data, and extracting time-domain, frequency-domain, and time-frequency-domain features. To achieve digital twin synchronization, monitor the synchronization error between the virtual model and the physical entity and adjust the model parameters, establish a data cache, and save historical data for trend analysis.

4. The fault diagnosis and self-healing control method for a flexible interconnection device in a distribution network as described in claim 3, characterized in that: The multi-scale feature fusion includes feature extraction through a convolutional neural network and calculation of attention mechanism weights to obtain key features; Perform feature importance assessment to quantify the contribution of extracted features to fault diagnosis; Implement multi-scale feature fusion to integrate information from different time and frequency scales.

5. The fault diagnosis and self-healing control method for a flexible interconnection device in a distribution network as described in claim 4, characterized in that: The fault classification process includes activating the fault mode recognition module, calculating the fault probability, and calculating the posterior probability of each fault type under a given observation based on Bayes' theorem. The support vector machine decision function is executed, and fault classification is performed using kernel functions and Lagrange multipliers. The neural network output is calculated, and the probability distribution of fault types is obtained through the softmax function. Calculate the diagnostic confidence level, assess the reliability of the diagnostic results, initiate expert system rule reasoning, calculate rule strength, fuzzy reasoning output, knowledge certainty factor and rule weight update respectively, and implement multi-model fusion decision-making.

6. The fault diagnosis and self-healing control method for a flexible interconnection device in a distribution network as described in claim 5, characterized in that: The self-healing control includes assessing the severity of the fault based on the fault diagnosis results and the weighted scores of all fault indicators. in, The severity of the fault, The weight of the i-th fault indicator. For the i-th fault indicator value, represents the total number of fault indicators, and i is the variable index; Select control action selection The optimal control strategy is determined based on the state value function, where, For optimal control action, The value function for performing action a in state s; The self-healing time window is calculated as follows: This includes fault detection time, decision-making time, and execution time, among which, Total self-healing time, For fault detection time, For decision-making time, For execution time; The control effect was evaluated, and the power recovery effect before and after self-healing was quantified. The control effect evaluation was as follows: in, For self-healing effect, Power after self-healing Power during a fault Normal power; Implement control actions, including protection device activation, load transfer, and parameter adjustment; monitor the self-healing process in real time; automatically switch to backup schemes when self-healing fails or is ineffective; and record self-healing process data.

7. The fault diagnosis and self-healing control method for a flexible interconnection device in a distribution network as described in claim 6, characterized in that: The startup fault early warning system estimates the probability of future faults, including... The fault early warning algorithm is activated to calculate the health index and assess the equipment health level by comprehensively evaluating the normalized status of various parameters. Degradation trend prediction is performed, and future state evolution is predicted based on the exponential smoothing algorithm; Calculate the probability prediction of failure, estimate the probability of future failures based on the exponential distribution model, dynamically adjust the warning threshold, and adaptively adjust the warning parameters based on the false alarm rate feedback. Different levels of early warning information are issued based on the risk level, including equipment status reminders, maintenance suggestions, and emergency warnings; Establish preventive maintenance plans, analyze the root causes of failures and propose improvement suggestions, update the failure knowledge base and experience data, and improve diagnostic models and control strategies; Regularly perform performance evaluations, calculate key performance indicators such as diagnostic accuracy, recall, F1 score, and self-healing success rate, analyze the diagnostic effectiveness of different fault types and adaptability to different operating scenarios, identify system performance bottlenecks and improvement opportunities, dynamically adjust algorithm parameters and optimize diagnostic strategies based on performance evaluation results, update deep learning models and expert system rule bases, integrate the latest fault characteristics and handling experience, and establish equipment lifecycle health management records.

8. A fault diagnosis and self-healing control system for a flexible interconnection device in a distribution network, using the fault diagnosis and self-healing control method for a flexible interconnection device in a distribution network as described in any one of claims 1 to 7, characterized in that, include: The system includes an initialization module, a data acquisition and preprocessing module, a multi-level fault feature extraction module, an intelligent fault diagnosis module, a self-healing control module, and a fault early warning module. Initialize the module, establish the system state vector, collect and weight the operating parameters of the distribution network, and initialize the fault diagnosis model parameters; The data acquisition and preprocessing module performs quality checks on the raw data based on the acquired operating parameters and dynamically adjusts the model parameters through digital twin synchronization and model parameter updates. The multi-level fault feature extraction module uses deep learning to extract multi-level fault features, quantifies the contribution of features to fault diagnosis, and performs multi-scale feature fusion. The intelligent fault diagnosis module activates the fault mode recognition module, calculates the fault probability to classify the fault, and evaluates the reliability of the diagnosis results. The self-healing control module assesses the severity of the fault based on the fault diagnosis results, selects the optimal control strategy, and performs self-healing control. The fault warning module initiates fault warnings to estimate the probability of future faults, issues warning information of different levels for preventive maintenance, and performs regular performance evaluations and optimizations.

9. 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 fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network as described in any one of claims 1 to 7.

10. 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 fault diagnosis and self-healing control method for a flexible interconnection device in a power distribution network as described in any one of claims 1 to 7.

Images