A fault analysis self-healing system of a photovoltaic circuit breaker in an intelligent power distribution cabinet

Abstract

This invention relates to the field of intelligent control and automation technology, and discloses a fault analysis and self-healing system for photovoltaic circuit breakers in intelligent distribution cabinets. The system includes a fault perception module, a diagnostic analysis module, a self-healing execution module, and a collaborative control module. The fault perception module collects multi-dimensional status signals in real time and marks features. The diagnostic analysis module deeply integrates and analyzes these signals, using intelligent algorithms to achieve accurate fault identification, location, assessment, and prediction, generating a diagnostic report and realizing comprehensive intelligent management of the circuit breaker's health status. The self-healing execution module automatically executes self-healing operations such as parameter adjustment, topology reconstruction, arc suppression, and isolation switching based on the diagnostic report and the self-healing strategy library. This allows for direct intervention in the early stages of a fault, restoring and maintaining equipment functionality, thus achieving a shift from passive protection to proactive intelligent self-healing.

Description

Technical Field

[0001] This invention relates to the field of intelligent control and automation technology, specifically to a fault analysis and self-healing system for photovoltaic circuit breakers in intelligent distribution cabinets. Background Technology

[0002] As a key hub between photovoltaic power generation systems and smart grids, distribution cabinets undertake the core functions of power collection, distribution, and protection. Photovoltaic circuit breakers, as key components within distribution cabinets to ensure the safe operation of circuits, are of paramount importance. During normal operation of the power system, photovoltaic circuit breakers must precisely control the opening and closing of circuits to ensure stable power transmission. When abnormal conditions such as overload or short circuit occur, faulty circuits must be disconnected in a timely manner to prevent the accident from escalating and to protect equipment and personnel safety.

[0003] Currently, in the operation and maintenance of photovoltaic circuit breakers in intelligent distribution cabinets, existing technologies typically rely on single electrical protection or periodic offline maintenance, lacking multi-dimensional, real-time, and comprehensive online perception and deep intelligent analysis of their operating status. This makes it difficult to achieve early and accurate identification of fault types, accurate location of fault roots, and effective prediction of fault evolution trends.

[0004] Therefore, a fault analysis and self-healing system for photovoltaic circuit breakers in intelligent distribution cabinets is proposed to solve the above problems. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a fault analysis and self-healing system for photovoltaic circuit breakers in intelligent distribution cabinets, which solves the problems mentioned in the background technology of difficulty in achieving early and accurate identification of fault types, accurate location of fault root causes, and effective prediction of fault evolution trends.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a fault analysis and self-healing system for photovoltaic circuit breakers in intelligent distribution cabinets, the system comprising a fault perception module, a diagnostic analysis module, a self-healing execution module, and a collaborative control module; The fault perception module is used to collect multi-dimensional status signals of the photovoltaic circuit breaker in real time during operation, including electrical parameter signals, mechanical characteristic signals, partial discharge signals and temperature field signals, to construct a comprehensive data set reflecting the overall health status of the circuit breaker, and to achieve preliminary extraction of fault features and anomaly marking. The diagnostic analysis module is used to perform deep fusion and intelligent analysis on the comprehensive data set. By combining signal processing, feature engineering and machine learning algorithms, it identifies fault types, locates fault root causes, assesses fault severity and predicts fault evolution trends, and generates a diagnostic report that includes specific fault modes, scope of impact and recommended handling strategies. The self-healing execution module is used to automatically and after confirmation perform corresponding physical self-healing operations based on the diagnostic report and the preset self-healing strategy library. The operations include parameter adaptive adjustment, internal topology flexible reconstruction, active suppression of fault arc, and online isolation and backup path switching of failed components, so as to restore and maintain the normal operation function of the circuit breaker. The collaborative control module is used to coordinate and schedule the timing operations and logical linkages of the fault perception module, diagnostic analysis module, and self-healing execution module, to verify and evaluate the diagnostic analysis results, and to dynamically optimize the self-healing strategy based on the self-healing execution effect. It also provides a visual human-computer interaction interface for system parameter configuration, real-time status monitoring, diagnostic process tracing, and self-healing record management.

[0007] Preferably, the fault perception module includes a multi-source sensing unit, a signal conditioning unit, and a feature extraction unit; The multi-source sensing unit consists of a high-precision current transformer, voltage sensor, vibration acceleration sensor, ultra-high frequency partial discharge sensor, infrared thermal imager and arc sensor array deployed at key locations of the circuit breaker, and is used to synchronously capture raw signals characterizing the electrical performance, mechanical action, insulation status and thermal characteristics of the circuit breaker. The signal conditioning unit includes an isolation amplifier circuit, an anti-aliasing filter circuit, and a high-speed analog-to-digital converter circuit, which are used to isolate and protect the original signal, reduce noise and filter it, and perform digital sampling. The feature extraction unit employs time-domain analysis, frequency-domain analysis, and time-frequency analysis algorithms to calculate various feature quantities such as effective value, peak value, harmonic content, vibration spectrum, discharge pulse sequence, and temperature gradient distribution from the preprocessed signal. Based on threshold comparison and statistical process control methods, it performs initial anomaly judgment and marking of the feature quantities.

[0008] Preferably, the multi-source sensing unit adopts a distributed layout and time synchronization technology to ensure that the timestamps of data collected by sensors at different physical locations are consistent. The signal conditioning unit has adaptive gain adjustment and common-mode rejection functions to adapt to wide-range fluctuations in photovoltaic current and strong electromagnetic interference environments. The feature extraction unit integrates wavelet packet transform and empirical mode decomposition algorithms to extract transient fault features from non-stationary signals, and performs dimensionality reduction and fusion of high-dimensional feature vectors through principal component analysis to generate a standardized feature set for subsequent diagnosis.

[0009] Preferably, the diagnostic analysis module includes a data fusion unit, an intelligent diagnostic unit, and a prediction and evaluation unit; The data fusion unit uses evidence theory to correlate and fuse multimodal feature information from different sensors, resolve feature conflicts, and form a consistent and complete description of the circuit breaker status. The intelligent diagnostic unit integrates an image recognition subunit based on a deep convolutional neural network, a sequence data analysis subunit based on a long short-term memory network, and a pattern recognition subunit based on a multi-class support vector machine. These subunits are used to process thermal images, time series signals, and structured feature data, respectively, to achieve accurate classification and location of various fault types such as external overheating, internal arcing, mechanical jamming, and insulation deterioration. The prediction and evaluation unit uses a survival analysis algorithm, combined with current fault characteristics and historical operating data, to assess the remaining effective lifespan and predict the time trajectory and risk probability of fault development.

[0010] Preferably, the intelligent diagnostic unit adopts a transfer learning strategy, uses a large amount of general electrical equipment fault data for model pre-training, and then fine-tunes it with a small amount of photovoltaic circuit breaker-specific data to overcome the problem of scarce fault samples in photovoltaic scenarios. The prediction and evaluation unit constructs a dynamic stress-intensity interference model that considers the randomness of photovoltaic output and the circuit breaker load rate, which is used to quantify the impact of operating condition fluctuations on the fault evolution rate and realize dynamic remaining lifetime prediction. The diagnostic analysis module also includes a diagnostic result credibility assessment mechanism. This mechanism assigns a credibility level to the final diagnostic conclusion by analyzing the consistency, feature significance, and model confidence score of the outputs of each diagnostic sub-unit. When the credibility is lower than a preset threshold, a review process is triggered and manual intervention is requested for confirmation.

[0011] Preferably, the self-healing execution module includes a parameter adjustment unit, a topology reconstruction unit, an arc suppression unit, and an isolation switching unit; The parameter adjustment unit uses controllable power electronic devices and intelligent adjustable components to finely adjust the tripping curve, protection setting and arc-extinguishing chamber arc-blowing pressure of the circuit breaker online, adapting to changes in operating conditions after a fault and compensating for performance deviations caused by component aging. The topology reconfiguration unit changes the connection method between the main contacts and parallel resistors and capacitors inside the circuit breaker through the built-in micro motor drive mechanism and solid-state switch array. When the main contact erosion degree is detected to exceed the preset threshold, the backup contact is activated to reconfigure the current path, homogenize the erosion and improve the breaking capacity. When the arc suppression unit detects signs of arcing, it injects a reverse current pulse into the fault point by triggering the pre-charged capacitor bank, controls the magnetic blow-out coil to generate a directional magnetic field, and completes the rapid forced extinguishing of the fault arc. After determining that a phase or component has suffered a permanent failure, the isolation switching unit controls the internal bypass switch to isolate the faulty part and simultaneously closes the preset redundant path to ensure that the non-faulty part continues to operate.

[0012] Preferably, the parameter adjustment unit is integrated with the digital trip unit of the circuit breaker, and supports dynamic modification of the threshold values ​​and time constants of long-delay, short-delay, and instantaneous protection through software commands; The topology reconfiguration unit adopts a modular design. Its actuator includes a miniature servo motor and linkage mechanism encapsulated in the circuit breaker's insulating housing, as well as a miniature switch array based on microelectromechanical technology. It is driven and controlled by a collaborative control module through optical fiber and an isolated power bus. The arc suppression unit includes a high-frequency pulse current generator and a fast magnetic control device, and its action response time is less than the time required for the fault arc to develop into a steady state. The isolation switching unit includes status detection and electrical interlocking logic, ensuring that no inrush current, circulating current, or secondary short circuit occurs during the process of isolating the faulty part and activating the redundant path.

[0013] Preferably, the collaborative control module includes a strategy scheduling unit, an effect evaluation unit, and a human-computer interaction unit; The strategy scheduling unit has a built-in task scheduler that combines time-triggered and event-triggered methods. It calls the corresponding self-healing execution logic according to the output priority of the diagnostic analysis module and manages resource conflicts and operation sequences when multiple tasks are executed concurrently. After the self-healing operation is executed, the effect evaluation unit continuously monitors the feedback data from the fault perception module and compares it with the baseline state before self-healing to quantitatively evaluate the effectiveness, efficiency, and impact on the overall system performance of the self-healing operation. Evaluation indicators include fault feature decay rate, function recovery time, and self-healing energy consumption. The human-computer interaction unit provides a graphical software interface and a hardware indicator panel. The interface supports three-dimensional fault location display, dynamic simulation of the self-healing process, and retrieval and comparison of historical case databases. The panel provides audible and visual alarms, self-healing status indicator lights, and emergency intervention buttons.

[0014] Preferably, the strategy scheduling unit adopts a decision framework based on a combination of rule engine and reinforcement learning. The rule engine processes known typical fault-self-healing mappings, while the reinforcement learning agent continuously optimizes self-healing strategies to cope with new and complex faults through interaction with the environment. The effect evaluation unit uses the following logic to quantify the immediate effect of a single self-healing action: Set up a set of key monitoring parameters, compare the root mean square error values ​​of the parameters in this set that deviate from the normal baseline within a specific time window before and after the completion of the self-healing operation, and calculate the percentage reduction of error as the instantaneous self-healing efficiency index. The collaborative control module also includes a knowledge update and evolution unit. This unit stores each complete fault diagnosis and self-healing process, including environmental conditions, fault characteristics, adopted strategies, and execution effects, into a structured case in the knowledge base, and uses case reasoning technology to optimize the diagnostic model parameters and self-healing strategy library.

[0015] Preferably, the system also includes a cloud collaboration and edge backup mechanism. The collaborative control module completes real-time diagnostic and self-healing tasks locally, while encrypting and uploading the desensitized running data, diagnostic reports and self-healing logs to the cloud analysis platform. The cloud platform aggregates operational data from multiple devices, utilizes computing resources for big data analysis and model retraining, and periodically distributes optimized diagnostic algorithm models and self-healing strategy packages to the collaborative control module on the edge side.

[0016] Compared with the prior art, the present invention provides a fault analysis and self-healing system for photovoltaic circuit breakers in intelligent distribution cabinets, which has the following beneficial effects: 1. In this invention, multi-dimensional status signals are collected in real time and their features are marked by the fault perception module. The diagnostic analysis module deeply integrates and analyzes these signals and uses intelligent algorithms to achieve accurate identification, location, assessment and prediction of faults, generate diagnostic reports, and realize comprehensive intelligent management of the health status of circuit breakers.

[0017] 2. In this invention, the self-healing execution module automatically performs self-healing operations such as parameter adjustment, topology reconstruction, arc suppression, and isolation switching based on the diagnostic report and the self-healing strategy library. It directly intervenes in the early stage of the fault, restores and maintains the equipment function, and realizes the transformation from passive protection to active intelligent self-healing.

[0018] 3. In this invention, the collaborative control module schedules the timing logic of each module, verifies and evaluates and optimizes the strategy, provides a visual human-machine interface, and with the help of cloud collaboration, edge backup and knowledge update unit, the system achieves reliable decision-making, efficient operation and maintenance and continuous self-evolution of performance. Attached Figure Description

[0019] Figure 1 This is a system architecture diagram of a fault analysis and self-healing system for photovoltaic circuit breakers in an intelligent power distribution cabinet according to the present invention. Figure 2 This is a module architecture diagram of a fault analysis and self-healing system for photovoltaic circuit breakers in an intelligent power distribution cabinet according to the present invention. Figure 3 This is a flowchart illustrating the operation steps of a fault analysis and self-healing system for photovoltaic circuit breakers in an intelligent distribution cabinet according to the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] For specific implementation examples, please refer to: Figure 1-3 A fault analysis and self-healing system for photovoltaic circuit breakers in an intelligent distribution cabinet, the system includes a fault perception module, a diagnostic analysis module, a self-healing execution module and a collaborative control module; The fault perception module is used to collect multi-dimensional status signals of the photovoltaic circuit breaker in real time during operation, including electrical parameter signals, mechanical characteristic signals, partial discharge signals and temperature field signals, to construct a comprehensive data set reflecting the overall health status of the circuit breaker, and to achieve preliminary extraction of fault characteristics and anomaly marking. The diagnostic analysis module is used to perform deep fusion and intelligent analysis on the comprehensive dataset. By combining signal processing, feature engineering and machine learning algorithms, it identifies fault types, locates the root causes of faults, assesses the severity of faults and predicts the evolution trend of faults, and generates a diagnostic report that includes specific fault modes, scope of impact and recommended handling strategies. The self-healing execution module is used to automatically and after confirmation perform corresponding physical self-healing operations based on the diagnostic report and the preset self-healing strategy library. The operations include parameter adaptive adjustment, internal topology flexible reconstruction, active suppression of fault arc, and online isolation and backup path switching of failed components, so as to restore and maintain the normal operation function of the circuit breaker. The collaborative control module coordinates and schedules the timing operations and logical linkages of the fault perception module, diagnostic analysis module, and self-healing execution module. It verifies and evaluates the diagnostic analysis results, dynamically optimizes the self-healing strategy based on the self-healing execution effect, and provides a visual human-machine interface for system parameter configuration, real-time status monitoring, diagnostic process tracing, and self-healing record management.

[0022] The fault perception module includes a multi-source sensing unit, a signal conditioning unit, and a feature extraction unit; The multi-source sensing unit consists of high-precision current transformers, voltage sensors, vibration acceleration sensors, ultra-high frequency partial discharge sensors, infrared thermal imagers, and arc light sensor arrays deployed at key locations of the circuit breaker. It is used to synchronously capture raw signals characterizing the electrical performance, mechanical action, insulation status, and thermal characteristics of the circuit breaker. The signal conditioning unit includes an isolation amplifier circuit, an anti-aliasing filter circuit, and a high-speed analog-to-digital converter circuit, which are used to isolate and protect the original signal, reduce noise and filter it, and perform digital sampling. The feature extraction unit uses time-domain analysis, frequency-domain analysis, and time-frequency analysis algorithms to calculate various feature quantities such as effective value, peak value, harmonic content, vibration spectrum, discharge pulse sequence, and temperature gradient distribution from the preprocessed signal. Based on threshold comparison and statistical process control methods, the feature quantities are initially judged and marked for anomalies. In practice, the feature extraction unit processes the conditioned multi-channel sensor signals in parallel; for the steady-state current and voltage signals, it performs time-domain analysis to calculate their effective value, peak value, and total harmonic distortion rate within one calculation cycle. The formula for calculating the total harmonic distortion rate is as follows: ; in Indicates the total harmonic distortion rate. This represents the effective value of the fundamental component. Indicates the first Effective value of the second harmonic component The highest order of the harmonics under consideration; For vibration acceleration signals, the kurtosis of their amplitude index is calculated to sensitively capture mechanical shocks. ; in The kurtosis of the signal amplitude distribution Represents the mathematical expectation. Represents the signal amplitude sequence. This represents the average amplitude of the signal. The standard deviation of the signal amplitude; For non-stationary partial discharge pulse signals, time-frequency analysis is performed, and continuous wavelet transform is used to calculate its wavelet coefficient matrix. The coefficient energy at a specific scale is extracted as a feature. ; in Indicates the signal at scale and time location The wavelet coefficients below, Represents the original time-domain signal. Represents the complex conjugate of the mother wavelet function. As a scale, For time location; All calculated characteristic quantities will be used to establish control limits based on statistical process control methods for initial anomaly detection; the mean of the characteristic quantities will be calculated from historical normal operation data. with standard deviation When real-time feature value satisfy If the condition is abnormal, it is marked as "abnormal"; meanwhile, for features such as temperature growth rate, it is directly compared with the empirical threshold and marked.

[0023] The multi-source sensing unit adopts a distributed layout and time synchronization technology to ensure that the timestamps of data collected by sensors at different physical locations are consistent. In practical implementation, the multi-source sensing unit adopts a distributed layout and time synchronization technology: the local frequency partial discharge sensor is embedded in the shielding cover of the arc-extinguishing chamber of each phase of the circuit breaker, the vibration acceleration sensor is installed at the operating mechanism shaft and contact support, and the infrared thermal imager is aimed at the moving and stationary contacts and the connecting busbar area; all sensors acquire signals through a data acquisition unit with a precision clock synchronization protocol to ensure microsecond-level time synchronization accuracy and provide a time reference for multi-physical quantity correlation analysis.

[0024] The signal conditioning unit has adaptive gain adjustment and common-mode rejection functions to adapt to wide-range fluctuations in photovoltaic current and strong electromagnetic interference environments; The signal conditioning unit features adaptive gain adjustment and common-mode rejection: the gain of its preamplifier circuit... Based on the amplitude of the input signal Dynamic adjustment, the logic is as follows Then improve ,when Then reduce ,in This serves as the ADC reference voltage, thereby fully utilizing the ADC's dynamic range.

[0025] The feature extraction unit integrates wavelet packet transform and empirical mode decomposition algorithms to extract transient fault features from non-stationary signals, and performs dimensionality reduction and fusion of high-dimensional feature vectors through principal component analysis to generate a standardized feature set for subsequent diagnosis. The formula for calculating the feature fusion weights is: ; in Indicates the first Features Weights in fusion Representation of features With fault category labels Mutual information between them The total number of features.

[0026] The diagnostic analysis module includes a data fusion unit, an intelligent diagnostic unit, and a predictive evaluation unit; The data fusion unit uses evidence theory to correlate and fuse multimodal feature information from different sensors, resolve feature conflicts, and form a consistent and complete description of the circuit breaker status. In practical implementation, the data fusion unit uses evidence theory to fuse multi-source information, treating the preliminary diagnostic results from current, vibration, temperature, and partial discharge sensors as basic probability allocation functions from different information sources. The current sensor's confidence level in the "overload" proposition is... The temperature sensor's confidence level regarding the "contact overheating" proposition is... ; Calculate the joint trust level using Dempster's combination rule: ; ; in and It is the basic probability assignment function for two independent sources of evidence. This indicates the effect of fusion on the proposition. The joint basic probability assignment, To identify propositions within the framework, To assess the conflict quality among evidence; this process integrates correlation and confidence levels to obtain a comprehensive failure confidence distribution; The intelligent diagnostic unit integrates an image recognition subunit based on a deep convolutional neural network, a sequence data analysis subunit based on a long short-term memory network, and a pattern recognition subunit based on a multi-class support vector machine. These subunits are used to process thermal images, time series signals, and structured feature data, respectively, to achieve accurate classification and location of various fault types such as external overheating, internal arcing, mechanical jamming, and insulation degradation. Deep convolutional neural networks are used to process infrared thermal images. Through convolutional and pooling layers, they automatically learn the spatial features of abnormal temperature areas to identify faults such as loose external connections and overheating of internal contacts. Long short-term memory networks are used to analyze the time sequence of current and vibration signals to capture the dynamic evolution patterns of faults such as arc reignition and mechanical jamming. Multi-class support vector machines receive fused structured feature vectors and solve for the hyperplane that maximizes the classification margin to achieve accurate classification and localization of fault types such as external overheating, internal arcing, mechanical jamming, and insulation degradation. The predictive assessment unit uses a survival analysis algorithm, combined with current fault characteristics and historical operating data, to assess the remaining effective life and predict the time trajectory and risk probability of fault development. The predictive assessment unit uses the proportional hazards model in the survival analysis algorithm to assess remaining lifespan, with the hazard function being: ; in Indicates that given a vector of covariates Under the conditions of time The risk function, Represents the benchmark risk function. This represents the coefficient vector related to the covariates. Represents a covariate vector; covariate vector It is composed of real-time monitored current fault characteristics and historical operating data, thereby mapping specific physical quantities to risk functions; Based on the proportional risk model, the equipment at time... The relationship between the survival function and the risk function is as follows: ; in For conditional survival functions, For conditional probability, Let be a random variable representing the equipment failure time. For at any time The conditional risk function; The equipment is in constant time The formula for predicting the expected value of remaining useful life is: ; in For a moment Using the expected remaining lifetime, The time variable is the inner multiple integral. For conditional survival functions, For at any time The conditional risk function; This formula obtains the survival probability from the current moment by integrating over all future time points. By combining the expected remaining time to failure with real-time updated covariates, dynamic prediction of the remaining effective lifetime can be achieved.

[0027] The intelligent diagnostic unit adopts a transfer learning strategy, using a large amount of general electrical equipment fault data for model pre-training, and then fine-tuning it with a small amount of photovoltaic circuit breaker-specific data to overcome the problem of scarce fault samples in photovoltaic scenarios. In practice, the intelligent diagnostic unit adopts a transfer learning strategy: First, a deep convolutional neural network model is pre-trained on a general dataset containing various electrical equipment faults, so that it learns a general fault feature representation; then, on a small-scale dataset of specific faults of photovoltaic circuit breakers, the parameters of the bottom feature extraction layer of the network are fixed, and only the top fully connected classification layer is fine-tuned. The optimization objective is to minimize the cross-entropy loss function on the target domain, so as to quickly adapt to the photovoltaic scenario. The prediction and evaluation unit constructs a dynamic stress-intensity interference model that considers the randomness of photovoltaic output and the circuit breaker load rate, which is used to quantify the impact of operating condition fluctuations on the fault evolution rate and realize dynamic remaining lifetime prediction. The dynamic stress-intensity interference model is constructed as follows: Define the "strength" of the circuit breaker contact resistance. Let be a degrading random variable that increases with the number of electrical wear cycles, and its distribution is as follows: Define "stress" The actual current flowing through the contacts, and its distribution Affected by both the randomness of photovoltaic power output and the circuit breaker load factor; therefore, at any given time... failure probability The probability that the stress exceeds the strength: ; in and These are the integral variables; based on this, the device at time... Remaining service life Through its survival function Make predictions; survival function Indicates the device at time The probability of it still functioning normally; the expected value of its dynamic remaining lifetime. The prediction formula is as follows: ; in This represents the expected value of the dynamic remaining lifetime. For at any time The failure probability function, For survival functions; This formula obtains the survival probability from the current moment by integrating over all future time points. The expected remaining time until failure; in practical applications, this is combined with real-time data to... By performing rolling calculations, dynamic remaining lifetime prediction can be achieved.

[0028] The diagnostic analysis module also has a diagnostic result credibility assessment mechanism. This mechanism assigns a credibility level to the final diagnostic conclusion by analyzing the consistency, feature significance, and model confidence score of the output of each diagnostic sub-unit. When the credibility is lower than the preset threshold, a review process is triggered and manual intervention is requested for confirmation. The confidence level is calculated using the following formula: ; in Represents conditional probability. Indicates the first Types of faults, This represents the input feature vector. For the diagnostic model to classify fault types The original output score, This represents the total number of fault types. The formula converts the original scores into a probability distribution, and the fault type corresponding to the maximum value is the diagnosis result, and its probability value is the confidence level.

[0029] The self-healing execution module includes a parameter adjustment unit, a topology reconstruction unit, an arc suppression unit, and an isolation switching unit; The parameter adjustment unit uses controllable power electronic devices and intelligent adjustable components to finely adjust the tripping curve, protection settings and arc-extinguishing chamber arc-blowing pressure of the circuit breaker online, adapting to changes in operating conditions after a fault and compensating for performance deviations caused by component aging. The topology reconfiguration unit uses a built-in micro motor drive mechanism and solid-state switch array to change the connection method between the main contacts and parallel resistors and capacitors inside the circuit breaker. When the main contact erosion degree exceeds the preset threshold, the backup contacts are activated to reconfigure the current path, homogenize erosion, and improve the breaking capacity. When the arc suppression unit detects signs of arcing, it injects a reverse current pulse into the fault point by triggering the pre-charged capacitor bank, controls the magnetic blow-out coil to generate a directional magnetic field, and completes the rapid forced extinguishing of the fault arc. After determining that a phase or component has suffered a permanent failure, the isolation switching unit controls the internal bypass switch to isolate the faulty part and simultaneously closes the preset redundant path to ensure that the non-faulty part continues to operate.

[0030] The parameter adjustment unit is integrated with the circuit breaker's digital trip unit, supporting dynamic modification of the threshold values ​​and time constants for long-delay, short-delay, and instantaneous protection via software commands; The topology reconfiguration unit adopts a modular design. Its actuators include a miniature servo motor and linkage mechanism encapsulated in the circuit breaker's insulating housing, as well as a miniature switch array based on microelectromechanical technology. It is driven and controlled by a collaborative control module through optical fiber and an isolated power bus. The arc suppression unit includes a high-frequency pulse current generator and a fast magnetic control device, and its action response time is less than the time required for the fault arc to develop into a steady state. The isolation switching unit includes status detection and electrical interlocking logic, ensuring that no inrush current, circulating current, or secondary short circuit occurs during the process of isolating the faulty part and activating the redundant path.

[0031] The collaborative control module includes a strategy scheduling unit, an effect evaluation unit, and a human-computer interaction unit; The strategy scheduling unit has a built-in task scheduler that combines time-triggered and event-triggered methods. It calls the corresponding self-healing execution logic based on the output priority of the diagnostic analysis module and manages resource conflicts and operation sequences when multiple tasks are executed concurrently. After the self-healing operation is executed, the effect evaluation unit continuously monitors the feedback data from the fault perception module and compares it with the baseline state before self-healing. It quantitatively evaluates the effectiveness, efficiency, and impact on the overall system performance of the self-healing operation. Evaluation indicators include fault feature decay rate, function recovery time, and self-healing energy consumption. In practice, after the self-healing operation is executed, the effect evaluation unit initiates a quantitative assessment of the effectiveness, efficiency, and impact on the overall system performance of the self-healing operation. Effectiveness is assessed by selecting key state parameters directly related to the fault, namely the temperature rise of the fault phase, and calculating the reduction percentage of the root mean square error of this parameter deviating from the normal baseline value within a time window before and after the completion of the self-healing operation. This reduction is defined as the effectiveness index. ; in Indicators representing the effectiveness of self-healing procedures. The function representing the calculation of the root mean square error. This indicates the sequence of fault-related parameters deviating from their normal baseline values ​​before the self-healing operation. This represents a sequence of the same parameter deviating from its normal baseline value after a self-healing operation. Efficiency assessment: Record the total time elapsed from the identification of fault characteristics to the completion of self-healing actions and the restoration of state parameters to the safe range. and compared with the preset maximum allowable processing time for the same type of fault. Compare; Assess the impact on overall system performance: Monitor the voltage dips of the system bus during and after the self-healing action. And the auxiliary energy consumed in the self-healing process The final output includes, , A quantitative assessment report on [the project / initiative]. The human-computer interaction unit provides a graphical software interface and a hardware indicator panel. The interface supports three-dimensional fault location display, dynamic simulation of the self-healing process, and retrieval and comparison of historical case databases. The panel provides audible and visual alarms, self-healing status indicator lights, and emergency intervention buttons.

[0032] The strategy scheduling unit adopts a decision framework based on a combination of rule engine and reinforcement learning. The rule engine processes known typical fault-self-healing mappings, while the reinforcement learning agent continuously optimizes self-healing strategies to deal with new and complex faults through interaction with the environment. In practice, the policy scheduling unit adopts a decision framework based on a combination of rule engine and reinforcement learning. For clear typical faults, the rule engine directly matches and executes the predefined "IF(fault type = a certain type AND severity = a certain level) THEN(execute action = a certain policy)" rule. For complex, novel, or compound faults, a reinforcement learning agent is activated. The agent will update the current system state. Mapping to self-healing actions After execution, the environment transitions to a new state. And give a reward The reward function is calculated by quantifying the output of the performance evaluation unit: ; in As a reward, , These are the weighting coefficients. The agent's goal is to maximize cumulative discount returns. It continuously updates its policy network parameters through temporal difference learning. : ; in Represents the approximate action value function Neural network parameters, Indicates the learning rate. Indicates a reward. Indicates the discount factor. and These represent the current state and the action being performed, respectively. Indicates the execution of an action The next state after transitioning to Indicates the use of target network parameters The estimated value of the optimal action in the next state. Indicates the current value pair parameter The gradient; The effectiveness evaluation unit uses the following logic to quantify the immediate effect of a single self-healing action: Set up a set of key monitoring parameters, compare the root mean square error values ​​of the parameters in this set that deviate from the normal baseline within a specific time window before and after the completion of the self-healing operation, and calculate the percentage reduction of error as the instantaneous self-healing efficiency index. The collaborative control module also has a knowledge update and evolution unit. This unit will store each complete fault diagnosis and self-healing process, including environmental conditions, fault characteristics, strategies adopted, and execution effects, into a structured case in the knowledge base. It will also use case reasoning technology to optimize the diagnostic model parameters and self-healing strategy library to achieve the system's self-learning and continuous improvement. The knowledge update and evolution unit utilizes case-based reasoning techniques: each complete fault handling process is encapsulated as a case. ,in For the problem description, As a solution, As a result, when a new fault occurs, the system retrieves the most similar historical case from the case library, and the similarity is calculated using cosine similarity in the feature space: ; in Indicates a new fault description With the historical case library Problem description of the case Similarity between them This represents the dot product of two eigenvectors; After retrieving the most similar cases, reuse and adjust their solutions, and store new successful cases in the database to achieve continuous knowledge accumulation.

[0033] The system also includes cloud collaboration and edge backup mechanisms. The collaborative control module completes real-time diagnostic and self-healing tasks locally, while encrypting and uploading the desensitized running data, diagnostic reports and self-healing logs to the cloud analysis platform. The cloud platform aggregates operational data from multiple devices, utilizes computing resources for big data analysis and model retraining, and periodically distributes optimized diagnostic algorithm models and self-healing strategy packages to the collaborative control module on the edge side, thereby achieving collective intelligent evolution and individual capability upgrades.

[0034] The operating steps of this system are as follows: First, the fault perception module collects multi-dimensional status signals such as electrical parameters, mechanical characteristics, partial discharge and temperature field signals during the operation of the photovoltaic circuit breaker in real time. After conditioning the original signals, feature quantities are extracted and anomalies are initially judged and marked to construct a comprehensive data set reflecting the overall health status of the circuit breaker.

[0035] Subsequently, the diagnostic analysis module performs in-depth fusion and intelligent analysis on the comprehensive dataset. Through its integrated data fusion, intelligent diagnosis and prediction assessment units, it identifies fault types, locates the root causes of faults, assesses the severity and predicts evolution trends, and finally generates a diagnostic report that includes specific fault modes, scope of impact and recommended handling strategies.

[0036] Next, the self-healing execution module automatically or after confirmation performs corresponding physical self-healing operations based on the diagnostic report and the preset self-healing strategy library, including parameter adaptive adjustment, internal topology flexible reconstruction, active suppression of fault arc, online isolation of failed components and switching of backup paths, so as to directly restore and maintain the normal operation function of the circuit breaker.

[0037] During this process, the collaborative control module is responsible for coordinating and scheduling the timing operations and logical linkages of the above modules, verifying and evaluating the diagnostic results, and dynamically optimizing the self-healing strategy based on the self-healing execution effect. At the same time, this module provides a system monitoring and management interface through the human-computer interaction unit, and with the help of cloud collaboration and edge backup mechanisms, realizes cloud analysis of data, model optimization and strategy updates, driving the system to complete self-learning and continuous improvement.

[0038] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0039] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A fault analysis and self-healing system for photovoltaic circuit breakers in intelligent distribution cabinets, characterized in that: The system includes a fault perception module, a diagnostic analysis module, a self-healing execution module, and a collaborative control module; The fault perception module is used to collect multi-dimensional status signals of the photovoltaic circuit breaker in real time during operation, including electrical parameter signals, mechanical characteristic signals, partial discharge signals and temperature field signals, to construct a comprehensive data set reflecting the overall health status of the circuit breaker, and to achieve preliminary extraction of fault features and anomaly marking. The diagnostic analysis module is used to perform deep fusion and intelligent analysis on the comprehensive data set. By combining signal processing, feature engineering and machine learning algorithms, it identifies fault types, locates fault root causes, assesses fault severity and predicts fault evolution trends, and generates a diagnostic report that includes specific fault modes, scope of impact and recommended handling strategies. The self-healing execution module is used to automatically and after confirmation perform corresponding physical self-healing operations based on the diagnostic report and the preset self-healing strategy library. The operations include parameter adaptive adjustment, internal topology flexible reconstruction, active suppression of fault arc, and online isolation and backup path switching of failed components, so as to restore and maintain the normal operation function of the circuit breaker. The collaborative control module is used to coordinate and schedule the timing operations and logical linkages of the fault perception module, diagnostic analysis module, and self-healing execution module, to verify and evaluate the diagnostic analysis results, and to dynamically optimize the self-healing strategy based on the self-healing execution effect. It also provides a visual human-computer interaction interface for system parameter configuration, real-time status monitoring, diagnostic process tracing, and self-healing record management.

2. The fault analysis and self-healing system for photovoltaic circuit breakers in an intelligent distribution cabinet according to claim 1, characterized in that: The fault perception module includes a multi-source sensing unit, a signal conditioning unit, and a feature extraction unit. The multi-source sensing unit consists of a high-precision current transformer, voltage sensor, vibration acceleration sensor, ultra-high frequency partial discharge sensor, infrared thermal imager and arc sensor array deployed at key locations of the circuit breaker, and is used to synchronously capture raw signals characterizing the electrical performance, mechanical action, insulation status and thermal characteristics of the circuit breaker. The signal conditioning unit includes an isolation amplifier circuit, an anti-aliasing filter circuit, and a high-speed analog-to-digital converter circuit, which are used to isolate and protect the original signal, reduce noise and filter it, and perform digital sampling. The feature extraction unit employs time-domain analysis, frequency-domain analysis, and time-frequency analysis algorithms to calculate various feature quantities such as effective value, peak value, harmonic content, vibration spectrum, discharge pulse sequence, and temperature gradient distribution from the preprocessed signal. Based on threshold comparison and statistical process control methods, it performs initial anomaly judgment and marking of the feature quantities.

3. The fault analysis and self-healing system for photovoltaic circuit breakers in an intelligent distribution cabinet according to claim 2, characterized in that: The multi-source sensing unit adopts a distributed layout and time synchronization technology to ensure that the timestamps of data collected by sensors at different physical locations are consistent. The signal conditioning unit has adaptive gain adjustment and common-mode rejection functions to adapt to wide-range fluctuations in photovoltaic current and strong electromagnetic interference environments. The feature extraction unit integrates wavelet packet transform and empirical mode decomposition algorithms to extract transient fault features from non-stationary signals, and performs dimensionality reduction and fusion of high-dimensional feature vectors through principal component analysis to generate a standardized feature set for subsequent diagnosis.

4. The fault analysis and self-healing system for photovoltaic circuit breakers in an intelligent distribution cabinet according to claim 1, characterized in that: The diagnostic analysis module includes a data fusion unit, an intelligent diagnostic unit, and a prediction and evaluation unit. The data fusion unit uses evidence theory to correlate and fuse multimodal feature information from different sensors, resolve feature conflicts, and form a consistent and complete description of the circuit breaker status. The intelligent diagnostic unit integrates an image recognition subunit based on a deep convolutional neural network, a sequence data analysis subunit based on a long short-term memory network, and a pattern recognition subunit based on a multi-class support vector machine. These subunits are used to process thermal images, time series signals, and structured feature data, respectively, to achieve accurate classification and location of various fault types such as external overheating, internal arcing, mechanical jamming, and insulation deterioration. The prediction and evaluation unit uses a survival analysis algorithm, combined with current fault characteristics and historical operating data, to assess the remaining effective lifespan and predict the time trajectory and risk probability of fault development.

5. The fault analysis and self-healing system for photovoltaic circuit breakers in an intelligent distribution cabinet according to claim 4, characterized in that: The intelligent diagnostic unit adopts a transfer learning strategy, uses a large amount of general electrical equipment fault data for model pre-training, and then fine-tunes it with a small amount of photovoltaic circuit breaker-specific data to overcome the problem of scarce fault samples in photovoltaic scenarios. The prediction and evaluation unit constructs a dynamic stress-intensity interference model that considers the randomness of photovoltaic output and the circuit breaker load rate, which is used to quantify the impact of operating condition fluctuations on the fault evolution rate and realize dynamic remaining lifetime prediction. The diagnostic analysis module also includes a diagnostic result credibility assessment mechanism. This mechanism assigns a credibility level to the final diagnostic conclusion by analyzing the consistency, feature significance, and model confidence score of the outputs of each diagnostic sub-unit. When the credibility is lower than a preset threshold, a review process is triggered and manual intervention is requested for confirmation.

6. The fault analysis and self-healing system for photovoltaic circuit breakers in an intelligent distribution cabinet according to claim 1, characterized in that: The self-healing execution module includes a parameter adjustment unit, a topology reconstruction unit, an arc suppression unit, and an isolation switching unit; The parameter adjustment unit uses controllable power electronic devices and intelligent adjustable components to finely adjust the tripping curve, protection setting and arc-extinguishing chamber arc-blowing pressure of the circuit breaker online, adapting to changes in operating conditions after a fault and compensating for performance deviations caused by component aging. The topology reconfiguration unit changes the connection method between the main contacts and parallel resistors and capacitors inside the circuit breaker through the built-in micro motor drive mechanism and solid-state switch array. When the main contact erosion degree is detected to exceed the preset threshold, the backup contact is activated to reconfigure the current path, homogenize the erosion and improve the breaking capacity. When the arc suppression unit detects signs of arcing, it injects a reverse current pulse into the fault point by triggering the pre-charged capacitor bank, controls the magnetic blow-out coil to generate a directional magnetic field, and completes the rapid forced extinguishing of the fault arc. After determining that a phase or component has suffered a permanent failure, the isolation switching unit controls the internal bypass switch to isolate the faulty part and simultaneously closes the preset redundant path to ensure that the non-faulty part continues to operate.

7. The fault analysis and self-healing system for photovoltaic circuit breakers in an intelligent distribution cabinet according to claim 6, characterized in that: The parameter adjustment unit is integrated with the circuit breaker's digital trip unit, and supports dynamic modification of the threshold values ​​and time constants for long-delay, short-delay, and instantaneous protection via software commands; The topology reconfiguration unit adopts a modular design. Its actuator includes a miniature servo motor and linkage mechanism encapsulated in the circuit breaker's insulating housing, as well as a miniature switch array based on microelectromechanical technology. It is driven and controlled by a collaborative control module through optical fiber and an isolated power bus. The arc suppression unit includes a high-frequency pulse current generator and a fast magnetic control device, and its action response time is less than the time required for the fault arc to develop into a steady state. The isolation switching unit includes status detection and electrical interlocking logic, ensuring that no inrush current, circulating current, or secondary short circuit occurs during the process of isolating the faulty part and activating the redundant path.

8. The fault analysis and self-healing system for photovoltaic circuit breakers in an intelligent distribution cabinet according to claim 1, characterized in that: The collaborative control module includes a strategy scheduling unit, an effect evaluation unit, and a human-computer interaction unit. The strategy scheduling unit has a built-in task scheduler that combines time-triggered and event-triggered methods. It calls the corresponding self-healing execution logic according to the output priority of the diagnostic analysis module and manages resource conflicts and operation sequences when multiple tasks are executed concurrently. After the self-healing operation is executed, the effect evaluation unit continuously monitors the feedback data from the fault perception module and compares it with the baseline state before self-healing to quantitatively evaluate the effectiveness, efficiency, and impact on the overall system performance of the self-healing operation. Evaluation indicators include fault feature decay rate, function recovery time, and self-healing energy consumption. The human-computer interaction unit provides a graphical software interface and a hardware indicator panel. The interface supports three-dimensional fault location display, dynamic simulation of the self-healing process, and retrieval and comparison of historical case databases. The panel provides audible and visual alarms, self-healing status indicator lights, and emergency intervention buttons.

9. The fault analysis and self-healing system for photovoltaic circuit breakers in an intelligent distribution cabinet according to claim 8, characterized in that: The policy scheduling unit adopts a decision framework based on a combination of rule engine and reinforcement learning. The rule engine processes known typical fault-self-healing mappings, while the reinforcement learning agent continuously optimizes self-healing strategies to deal with new and complex faults through interaction with the environment. The effect evaluation unit uses the following logic to quantify the immediate effect of a single self-healing action: Set up a set of key monitoring parameters, compare the root mean square error values ​​of the parameters in this set that deviate from the normal baseline within a specific time window before and after the completion of the self-healing operation, and calculate the percentage reduction of error as the instantaneous self-healing efficiency index. The collaborative control module also includes a knowledge update and evolution unit. This unit stores each complete fault diagnosis and self-healing process, including environmental conditions, fault characteristics, adopted strategies, and execution effects, into a structured case in the knowledge base, and uses case reasoning technology to optimize the diagnostic model parameters and self-healing strategy library.

10. The fault analysis and self-healing system for photovoltaic circuit breakers in an intelligent distribution cabinet according to claim 1, characterized in that: The system also includes a cloud collaboration and edge backup mechanism. The collaborative control module completes real-time diagnostic and self-healing tasks locally, while encrypting and uploading the desensitized running data, diagnostic reports and self-healing logs to the cloud analysis platform. The cloud platform aggregates operational data from multiple devices, utilizes computing resources for big data analysis and model retraining, and periodically distributes optimized diagnostic algorithm models and self-healing strategy packages to the collaborative control module on the edge side.

Images