Charging fault real-time diagnosis system based on edge calculation

Abstract

The invention provides a charging fault real-time diagnosis system based on edge calculation, and relates to the technical field of charging fault diagnosis. The method comprises the following steps: constructing a three-dimensional space model of a target charging area, and dividing the target area into M monitoring sub-areas; collecting charging fault data of each monitoring sub-region; preprocessing the charging fault data; constructing a fault risk assessment model by using an echo state network, and optimizing model hyper-parameters by using a pollen propagation algorithm; deploying a fixed edge node in a high-risk area in the fault risk level of each monitoring sub-area for continuous monitoring, planning an inspection path of a mobile edge node, and executing monitoring according to a dynamic period; receiving monitoring data of fixed and mobile edge nodes in real time, and constructing a fault prediction model based on a support vector machine to obtain a fault probability value; and the fault probability value is compared with a preset multi-level early warning threshold value, the fault level is judged, and the early warning information is output, so that the safety and the reliability of the charging process are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of charging fault diagnosis, and in particular to a real-time charging fault diagnosis system based on edge computing. Background Art

[0002] With the increasing popularity of two-wheeled electric vehicles, charging safety issues are becoming increasingly prominent. Existing charging fault monitoring methods often rely on centralized cloud-based analysis, which suffers from significant data transmission latency and high network dependency, making it difficult to meet the demands of real-time fault response. Traditional fault detection relies solely on a single voltage or current threshold, making it susceptible to interference and false alarms, and unable to accurately identify the fault type. Furthermore, most charging stations lack proactive early warning mechanisms, often responding passively only after a fault occurs, which can easily lead to safety incidents such as fires.

[0003] The existing charging process can present a variety of fault types, including overcharging, short circuits, poor contact, and battery overheating. Each fault type has its own unique development process and manifestation characteristics, often accompanied by a sustained, abnormal increase in battery voltage, exceeding the normal charging cutoff voltage range, and the rate of increase is relatively slow. Short circuit faults, on the other hand, typically cause a sudden, sharp increase in current, with very rapid changes. Traditional systems employ simple data processing techniques, failing to fully exploit the value of data. Model parameters are often set based on empirical experience, making them difficult to adapt to complex and changing charging scenarios, resulting in poor fault assessment and prediction accuracy. Summary of the Invention

[0004] The present invention provides a real-time charging fault diagnosis system based on edge computing, which is used to solve the defects of warning timeliness and accuracy caused by the single warning threshold in the existing technology.

[0005] In one aspect, the present invention provides a real-time charging fault diagnosis system based on edge computing, comprising: Region division module: Construct a three-dimensional spatial model of the target charging area, mark the distribution density of charging equipment and user usage frequency, and divide the target area into M monitoring sub-areas.

[0006] Edge perception module: collects charging fault data in each monitoring sub-area.

[0007] The data preprocessing module is used to preprocess the charging fault data to obtain preprocessed charging fault data.

[0008] The risk grading module is used to build a fault risk assessment model using the echo state network, optimize the model hyperparameters using the pollen propagation algorithm, and input the preprocessed charging fault data into the fault risk assessment model to obtain the fault risk level of each monitoring sub-area.

[0009] Dynamic Monitoring Module: Deploy fixed edge nodes for continuous monitoring in medium- and high-risk areas within each monitoring sub-area. Optimize the location and number of edge nodes based on their monitoring radius. Combine medium- and low-risk areas not covered by fixed nodes into mobile monitoring areas. Plan patrol routes for mobile edge nodes and perform monitoring on a dynamic basis.

[0010] Intelligent early warning module: Receives monitoring data from fixed and mobile edge nodes in real time, builds a fault prediction model based on support vector machines, and obtains the fault probability value.

[0011] The fault warning module is used to compare the fault probability value with the preset multi-level warning threshold, determine the fault level, and output warning information.

[0012] According to the real-time charging fault diagnosis system based on edge computing provided by the present invention, the charging fault data includes voltage data, current data, battery data, and device status data.

[0013] Voltage data includes: real-time voltage value, voltage fluctuation amplitude, and voltage harmonic content.

[0014] Current data includes: real-time current value, current change rate, and current harmonic distortion rate.

[0015] Battery data includes: battery surface temperature, battery temperature gradient, battery state of charge, and battery health status.

[0016] Device status data includes: plug-in and plug-out times, contact resistance, operating mode, and fault indicator light status.

[0017] According to the real-time charging fault diagnosis system based on edge computing provided by the present invention, the data preprocessing module includes: The noise filtering unit is used to eliminate power frequency interference and high-frequency noise using an adaptive LMS algorithm.

[0018] The normalization processing unit is used to normalize the voltage and current signals to the [0, 1] interval.

[0019] The outlier correction unit is used to identify and replace abnormal data points using the 3σ criterion.

[0020] The feature extraction unit is used to extract time domain features and frequency domain features from the charging data.

[0021] According to the edge computing-based real-time charging fault diagnosis system provided by the present invention, the specific steps of using the echo state network to build a fault risk assessment model are as follows: Determine the network topology and reserve pool parameter configuration.

[0022] The reservoir state is calculated based on the determined network topology and reservoir parameter configuration.

[0023] The training sequence is forward propagated according to the state of the reserve pool, the state of the reserve pool x(n) at all time steps is recorded, and the output is the fault risk level label y(n).

[0024] Construct the reservoir state matrix X and output matrix Y based on the reservoir state x(n) at all time steps, and use Tikhonov regularization to calculate the optimal output weights.

[0025] The optimal output weight W out It was integrated into the model, and the generalization ability of the model was evaluated using five-fold cross validation to obtain the fault risk assessment model.

[0026] According to the edge computing-based charging fault real-time diagnosis system provided by the present invention, the specific steps of using the pollen propagation algorithm to optimize the model hyperparameters are as follows: Identify the echo state network hyperparameters that need to be optimized and set the value range of each parameter.

[0027] M pollen individuals are randomly generated, each of which is a combination of a set of ESN hyperparameters to form the initial population.

[0028] Using the cross-validation accuracy as the evaluation indicator, the ESN model corresponding to each pollen individual was trained and validated, and its fitness value was calculated.

[0029] Iterate over each pollen individual.

[0030] Compare the fitness values ​​of all pollen individuals and retain the optimal solution.

[0031] Determine whether the maximum number of iterations has been reached or the fitness value has converged. If so, stop the optimization process and output the global optimal solution as the final hyperparameter. Otherwise, continue iterating.

[0032] According to the real-time charging fault diagnosis system based on edge computing provided by the present invention, the specific steps of the iterative process for each pollen individual are: For each pollen individual, a random number is generated. If the random number is less than the probability p, global pollination is performed and the solution is updated according to the Lévy flight. Otherwise, local pollination is performed and updated between adjacent solutions.

[0033] Determine whether the updated solution exceeds the parameter value range, and if so, project it back into the boundary.

[0034] According to the real-time charging fault diagnosis system based on edge computing provided by the present invention, the formula of the Levy flight update solution is expressed as: Where L is the step size of the Levy distribution, is the position of the global optimal solution in the entire population at the tth iteration, is the position of the i-th solution at the t-th iteration, represents the updated position of the i-th solution at the t+1-th iteration.

[0035] According to the real-time charging fault diagnosis system based on edge computing provided by the present invention, the specific steps of the dynamic monitoring module to optimize the layout position of fixed edge nodes are as follows: Set the center coordinate set of the high-risk area to Q and the edge node monitoring radius to R.

[0036] Initialize the empty set D and the uncovered area set V.

[0037] Select region q from V i , calculate q i The maximum straight-line distance to all other regions in V.

[0038] Judge d imax Is it less than 2R? If so, set q i Add D and remove all distances q from V i The area is smaller than R.

[0039] Repeatedly optimize the fixed edge node layout until V is empty, and the output set D is the minimum number of edge nodes required.

[0040] According to the real-time charging fault diagnosis system based on edge computing provided by the present invention, in the dynamic monitoring module, the specific steps of path planning in the mobile monitoring area are as follows: The coordinate set of n sampling points that collect the boundary of the mobile monitoring area is S.

[0041] Combining the distribution density of medium- and low-risk sub-regions and the sampling point coordinate set S, the regional boundary curve is fitted using the cubic spline function.

[0042] The minimum rectangular range that can completely cover the area is determined based on the boundary curve. Uniform grid points are generated within the rectangular range with the effective monitoring width of the mobile edge node as the interval. Points located within the boundary curve and corresponding to the centers of medium and low-risk sub-areas are screened out to form a set of must-pass points.

[0043] A genetic algorithm is used to optimize the path of the necessary points, with the goal of minimizing the total length of the path and covering all medium- and low-risk sub-areas. The path order is iteratively optimized, and a dynamic weight factor is introduced to give higher weights to medium- and low-risk sub-areas that have recently experienced minor anomalies, so that they are covered first in the inspection path.

[0044] Adjust the inspection cycle according to the season and time period to ensure that mobile monitoring can provide comprehensive coverage and focus on potential risk points.

[0045] According to the edge computing-based real-time charging fault diagnosis system provided by the present invention, the fault prediction model construction of the intelligent early warning module includes: The extracted time domain features and frequency domain features are formed into a feature vector set.

[0046] The feature vector set is divided into a training set and a validation set. The support vector machine model is constructed using the training set. The radial basis kernel function is selected as the kernel function. The penalty coefficient C and kernel function parameter γ of the support vector machine model are optimized by the grid search method.

[0047] The performance of the optimized support vector machine model is evaluated using the validation set.

[0048] The support vector machine model that passes the evaluation is the fault prediction model. The feature vector of the real-time monitoring data is input to obtain the fault probability value.

[0049] The present invention provides a real-time charging fault diagnosis system based on edge computing. By using an echo state network combined with a pollen propagation algorithm to optimize hyperparameters, a fault risk assessment model is constructed. The fault risk level of each monitoring sub-area is derived based on preprocessed data. The system achieves the following beneficial effects: the real-time charging fault diagnosis system constructs a three-dimensional spatial model of the target charging area, clarifies the distribution density of charging equipment and user usage frequency, and divides the monitoring sub-areas. The edge perception module collects charging fault data from each sub-area. The echo state network is combined with a pollen propagation algorithm to optimize hyperparameters to construct a fault risk assessment model. The fault risk level of each monitoring sub-area is derived based on the preprocessed data. Fixed edge nodes are then deployed in high-risk areas for continuous monitoring and their deployment is optimized based on different risk levels. Medium and low-risk areas are merged into mobile monitoring areas, and inspection routes are planned for dynamic periodic monitoring. The intelligent early warning module then constructs a support vector machine fault prediction model based on the monitoring data of fixed and mobile edge nodes to obtain a fault probability value. Finally, the fault early warning module compares the fault probability value with a preset multi-level early warning threshold to determine the fault level and output early warning information, thereby achieving comprehensive and accurate monitoring of the charging area, effectively capturing potential fault risks in different areas, and improving the accuracy of fault diagnosis. Monitoring resources can be reasonably allocated according to risk levels, and the layout and monitoring methods of fixed and mobile edge nodes can be optimized. This reduces resource costs while ensuring monitoring effectiveness. By using multi-level warning thresholds and comparisons of fault probability values, flexible and accurate graded warnings are achieved, avoiding the false alarms and missed alarms that are prone to occur with a single threshold in traditional systems. This greatly improves the timeliness and accuracy of warnings, and forms a complete and efficient real-time diagnosis system for charging faults from data collection, risk assessment to real-time warnings, effectively ensuring the safety and reliability of the charging process.

[0050] The combination of an echo state network (ESN) and a pollen propagation algorithm has been applied to the real-time diagnosis of charging faults in two-wheeled electric vehicles. By collaboratively optimizing the model structure and parameters, the system significantly improves its ability to identify complex faults and its early warning efficiency. As a recurrent neural network, the ESN, with its dynamic reservoir characteristics, effectively captures the nonlinear temporal characteristics of the charging process and is particularly adept at handling fault evolution processes with memory effects. The performance of traditional ESNs is highly dependent on hyperparameter settings, and improper parameter selection can easily lead to overfitting or underfitting. The introduction of the pollen propagation algorithm addresses this pain point: its global search capability, based on the biological pollen propagation mechanism, can quickly locate the optimal parameter combination in a high-dimensional parameter space, enabling the ESN model to maintain good generalization even with limited training data. By optimizing the ESN's reservoir dynamics, the model's sensitivity to weak fault signals is enhanced, improving the fault detection rate by approximately 15%-20%. The FPA's Levy flight mechanism enables the algorithm to escape local optima, effectively avoiding the local minima that plague traditional gradient descent methods, and reducing the model's false alarm rate by over 30% under complex operating conditions. The hybrid model more accurately models the temporal dependencies of charging data, enabling prediction of potential faults 3-5 minutes in advance, approximately 50% earlier than traditional machine learning methods. Furthermore, FPA's iterative optimization process features adaptive weight adjustment, dynamically balancing exploration and development based on real-time data characteristics, enabling the model to maintain stable performance despite seasonal environmental changes. This algorithmic synergy also reduces reliance on large-scale annotated data, enabling high-precision fault classification through small-sample learning. This is particularly well-suited for the data sparsity and fault diversity inherent in two-wheeled electric vehicle charging scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0052] Figure 1 1 is a module diagram of a real-time charging fault diagnosis system based on edge computing provided by an embodiment of the present invention; Figure 2 It is a flow chart of the present invention using the pollen propagation algorithm to optimize model hyperparameters. DETAILED DESCRIPTION

[0053] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0054] The following combination Figure 1-Figure 2 The present invention describes a real-time charging fault diagnosis system based on edge computing.

[0055] like Figure 1-Figure 2 As shown, the embodiment of the present invention provides a real-time charging fault diagnosis system based on edge computing, and the system includes: Region division module: Construct a three-dimensional spatial model of the target charging area, mark the distribution density of charging equipment and user usage frequency, and divide the target area into M monitoring sub-areas.

[0056] Using LiDAR or 3D scanning technology, the target charging area is scanned in all directions, collecting millimeter-level point cloud data containing ground elevation, building outlines, charging station locations, and coordinates of obstacles such as walls and columns. This point cloud data is imported into a geographic information system or 3D modeling software, and through point cloud registration, noise reduction, and gridding, a 3D digital model containing spatial coordinates (X, Y, Z) is generated.

[0057] The density of charging devices and user usage frequency were plotted. In the 3D model, the number of charging devices within each unit (10 square meters) was calculated, and the density levels were plotted using a heat map. The central area of ​​a community carport might exhibit a high density due to the concentration of devices, while the peripheral areas might exhibit a low density.

[0058] User usage frequency: Based on the historical usage records of charging devices, the model annotates the usage frequency of each device. Furthermore, through video surveillance or user movement trajectory data, paths with high usage frequency are annotated to reflect the intensity of human activity in the area.

[0059] The target area is divided into M monitoring sub-areas. Based on the spatial structure, equipment density, and usage frequency of the 3D model, the adaptive grid division method is used to determine the M monitoring sub-areas. The specific steps are as follows: Based on the physical boundaries in the 3D model, ensure that the sub-area does not cross obstacles to avoid obstruction of the monitoring range.

[0060] High-density equipment areas are divided into smaller sub-areas to ensure that every device is within the monitoring range. Low-density equipment areas are divided into larger sub-areas to reduce redundant divisions.

[0061] Include high-frequency equipment and the surrounding 1-meter range into the same sub-area, giving priority to ensuring monitoring accuracy in high-risk activity areas.

[0062] The M value is dynamically adjusted according to the total area of ​​the region to ensure that the shape of each sub-region is close to a rectangle and the long side is no more than twice the short side, which facilitates the monitoring coverage calculation of subsequent edge nodes.

[0063] Edge sensing module: Collects charging fault data in each monitoring sub-area. Charging fault data includes voltage data, current data, battery data, and device status data.

[0064] Voltage data includes: real-time voltage value, voltage fluctuation amplitude, and voltage harmonic content.

[0065] Current data includes: real-time current value, current change rate, and current harmonic distortion rate.

[0066] Battery data includes: battery surface temperature, battery temperature gradient, battery state of charge, and battery health status.

[0067] Device status data includes: plug-in and plug-out times, contact resistance, operating mode, and fault indicator light status.

[0068] The data preprocessing module is used to preprocess the charging fault data to obtain preprocessed charging fault data. The data preprocessing module includes: The noise filtering unit is used to eliminate 50Hz power frequency interference and high-frequency noise using an adaptive LMS algorithm.

[0069] The normalization processing unit is used to normalize the voltage and current signals to the range [0, 1] to eliminate dimensional differences. The adaptive filter is used to remove high-frequency interference from the power grid, and the Z-score is used to normalize the data dimensions. The calculation formula is: In the formula, μ is the signal mean, σ is the signal standard deviation, and the standardized data is more conducive to feature extraction.

[0070] The outlier correction unit is used to identify and replace abnormal data points using the 3σ criterion.

[0071] The feature extraction unit is used to extract time domain features and frequency domain features from the charging data.

[0072] Time domain feature extraction includes the charging current fluctuation amplitude, which reflects contact stability, and the temperature change rate, calculated at an interval of 1 second, to monitor the battery temperature rise trend.

[0073] Frequency domain feature extraction includes: performing wavelet transform on the current signal and decomposing it to obtain high-frequency components above 1kHz, which are used to identify the spark discharge characteristics caused by poor contact.

[0074] The risk grading module is used to build a fault risk assessment model using the echo state network, optimize the model hyperparameters using the pollen propagation algorithm, and obtain the fault risk level of each monitoring sub-area.

[0075] The specific steps of using echo state network to build a fault risk assessment model are: Determine the network topology and reservoir parameter configuration. The network topology includes: input layer dimensions, which are equal to the number of extracted features; the number of reservoir neurons, set to 100-500; and the output layer dimensions, which are equal to the number of fault risk levels. Reservoir parameter configuration includes sparse connectivity ratios set to 0.1-0.3, spectral radius set to 0.8-1.2, input scaling factor set to 0.1-1.0, and leakage rate set to 0.2-0.8.

[0076] Risk levels include: low risk level, medium risk level, and high risk level.

[0077] Low-risk level: Within the monitoring sub-area corresponding to this level, charging equipment operating parameters remain within normal ranges for a long period of time, with a historical failure rate of less than 0.5 events per month and no potential failure characteristics. The predicted failure probability, as predicted by the echo state network model, is less than 30%, indicating a very low probability of short-term charging failure. Only regular basic inspections are required, with no additional intervention required.

[0078] Medium risk: This level corresponds to monitoring sub-areas with minor anomalies, such as occasional increases in charging port contact resistance, voltage fluctuations exceeding the normal range by 10%-20%, and temperature fluctuations near the upper limit of the normal threshold. A historical failure rate of 0.5-2 times per month is observed. The model predicts a failure probability of 30%-70%, indicating a potential failure risk but not yet a serious threat. Increased monitoring frequency is required, with a focus on tracking the trend of abnormal parameter changes and, if necessary, preventive maintenance.

[0079] High-risk level: This level corresponds to monitoring sub-areas with significant abnormal characteristics, including sudden surges and dips in charging current, temperatures exceeding safety thresholds, sparks or odors on the interface, and a historical failure rate greater than two per month. A model-predicted failure probability greater than 70% indicates a high risk of a serious failure in the near term. Immediately initiate enhanced monitoring, trigger an early warning mechanism, and, if necessary, suspend charging services in the area until the potential fault is resolved.

[0080] The reserve pool status is calculated based on the determined network topology and reserve pool parameter configuration. The reserve pool status calculation formula is: Where x(n) is the state vector of the reservoir, u(n) is the input vector, and W inis the input weight matrix, W is the internal connection matrix of the reservoir, and α is the leakage rate.

[0081] The reservoir state vector is applied to the forward propagation of the training sequence, and the reservoir state x(n) of all time steps is recorded. The corresponding target output is the fault risk level label y(n).

[0082] The reservoir state matrix X is constructed using the reservoir state x(n) recorded at all time steps. Combined with the corresponding target output matrix Y, the optimal output weight is calculated using Tikhonov regularization. The formula is expressed as: Where X is the reservoir state matrix of all time steps, Y is the corresponding target output matrix, and λ is the regularization parameter.

[0083] The calculated optimal output weight W out It was integrated into the model, and the generalization ability of the model was evaluated using five-fold cross validation to obtain the fault risk assessment model.

[0084] The specific steps for optimizing model hyperparameters using the pollen dispersal algorithm are: Identify the echo state network hyperparameters that need to be optimized and set the value range of each parameter.

[0085] M pollen individuals are randomly generated, each of which is a combination of a set of ESN hyperparameters to form the initial population.

[0086] Using the cross-validation accuracy as the evaluation indicator, the ESN model corresponding to each pollen individual was trained and validated, and its fitness value was calculated.

[0087] For each pollen individual, a random number between 0 and 1 is generated. If it is less than the probability p, global pollination is performed, otherwise local pollination is performed.

[0088] Simulating global pollination, the updated solution formula according to Levy flight is expressed as: Where L is the step size of the Levy distribution, i.e. the parameter update amplitude during global pollination, x bestt is the current global optimal solution.

[0089] Simulating local pollination, the update formula between adjacent solutions is expressed as: Where j and k are different individuals selected randomly, and ε is a random number between 0 and 1.

[0090] Determine whether the updated solution exceeds the parameter value range, and if so, project it back into the boundary.

[0091] Compare the fitness values ​​of all pollen individuals and retain the optimal solution.

[0092] Determine whether the maximum number of iterations has been reached or the fitness value has converged. If so, stop the optimization process and output the global optimal solution as the final hyperparameter. Otherwise, continue iterating.

[0093] Dynamic Monitoring Module: Deploy fixed edge nodes for continuous monitoring in medium- and high-risk areas within each monitoring sub-area. Optimize the location and number of edge nodes based on their monitoring radius. Combine medium- and low-risk areas not covered by fixed nodes into mobile monitoring areas. Plan patrol routes for mobile edge nodes and perform monitoring on a dynamic basis.

[0094] Assume that there are K high-risk sub-areas in the target area, and the center coordinates of each area are marked as q1, q2, ..., q K , the set is represented as , suppose that the effective monitoring range of each fixed edge node is a circular area with a radius of R.

[0095] The initialized empty set D is used to store the final edge node layout positions. The uncovered area set V is initially equal to the high-risk area set Q. As the optimization process progresses, the covered areas are gradually removed.

[0096] Randomly select a region q from the currently uncovered region set V i , calculate q i The straight-line distance d to all other regions in V (j=i) ij , and find the maximum value d imax The maximum distance calculation formula for other areas is expressed as: Judge d imax Is it less than 2R? i The circular area with radius R as the center can cover all other areas in V. If so, q i Add D and remove all distances q from V i Areas smaller than R are already covered.

[0097] Judge d imax Is it greater than 2R, indicating that there is a distance q in V i Too far away, only relying on q i If coverage is not possible, reselect another area and recalculate the maximum distance of the uncovered area.

[0098] Repeat the above steps, each time selecting a new candidate point from the remaining uncovered area V, calculating the maximum distance and determining whether to set it as an edge node, until V is empty.

[0099] All q contained in the set D i This is the optimal fixed edge node layout location, and |D| is the minimum number of edge nodes required.

[0100] Intelligent early warning module: Receives monitoring data from fixed and mobile edge nodes in real time, builds a fault prediction model based on support vector machines, outputs fault probability values, and triggers graded early warnings if the value exceeds the preset threshold.

[0101] The fault prediction model construction of the intelligent early warning module includes: The extracted time domain features and frequency domain features are formed into a feature vector set.

[0102] The feature vector set is divided into a training set and a validation set. The support vector machine model is constructed using the training set. The radial basis kernel function is selected as the kernel function. The penalty coefficient C and kernel function parameter γ of the support vector machine model are optimized by the grid search method.

[0103] The performance of the optimized support vector machine model is evaluated using the validation set.

[0104] The support vector machine model that passes the evaluation is the fault prediction model. The feature vector of the real-time monitoring data is input to obtain the fault probability value.

[0105] The fault warning module is used to compare the fault probability value with the preset multi-level warning threshold, determine the fault level, and output warning information.

[0106] In summary, this embodiment provides a real-time charging fault diagnosis system based on edge computing. This system constructs a three-dimensional spatial model of the target charging area, identifies the distribution density of charging devices and user usage frequency, and divides the area into monitoring subareas. Based on this model, the edge sensing module collects charging fault data from each subarea. This data is then preprocessed by the data preprocessing module. An echo state network is then combined with a pollen propagation algorithm to optimize hyperparameters and construct a fault risk assessment model. The preprocessed data is used to determine the fault risk level of each monitoring subarea. Fixed edge nodes are then deployed in high-risk areas for continuous monitoring and their deployment is optimized based on the risk level. Medium- and low-risk areas are merged into mobile monitoring areas, and inspection routes are planned for dynamic periodic monitoring. The intelligent early warning module then constructs a support vector machine fault prediction model based on the monitoring data from fixed and mobile edge nodes to obtain a fault probability value. Finally, the fault early warning module compares the fault probability value with preset multi-level warning thresholds to determine the fault level and output warning information. This achieves comprehensive and accurate monitoring of the charging area, effectively capturing potential fault risks in different areas and improving the accuracy of fault diagnosis. Monitoring resources can be rationally allocated based on risk levels, optimizing the layout and monitoring methods of fixed and mobile edge nodes, reducing resource costs while ensuring monitoring effectiveness. By comparing multi-level warning thresholds and fault probability values, flexible and accurate graded warnings are achieved, avoiding the false alarms and missed alarms that are common in traditional systems with a single threshold. This significantly improves the timeliness and accuracy of warnings, forming a complete and efficient real-time diagnosis system for charging faults, from data collection and risk assessment to real-time warnings, effectively ensuring the safety and reliability of the charging process.

[0107] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion 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 ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods of each embodiment or certain portions of the embodiments.

[0108] Finally, it should be noted that the above embodiments are intended only to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art will appreciate that modifications may be made to the technical solutions described in the aforementioned embodiments, or that some of the technical features may be replaced with equivalents. Such modifications or replacements do not deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. Real-time diagnosis system for charging faults based on edge computing, characterized by: include: The area division module is used to build a three-dimensional spatial model of the target charging area, annotate the distribution density of charging equipment and user usage frequency, and divide the target area into M monitoring sub-areas; Edge sensing module, used to collect charging fault data in each monitoring sub-area; a data preprocessing module, configured to preprocess the charging fault data to obtain preprocessed charging fault data; a risk grading module, configured to construct a fault risk assessment model using an echo state network, optimize model hyperparameters using a pollen propagation algorithm, and input the pre-processed charging fault data into the fault risk assessment model to obtain a fault risk level for each monitoring sub-area; A dynamic monitoring module is used to deploy fixed edge nodes for continuous monitoring of medium- and high-risk areas of each monitoring sub-area, optimizing the location and number of edge nodes based on their monitoring radius; merging medium- and low-risk areas not covered by fixed nodes into mobile monitoring areas, planning inspection paths for mobile edge nodes, and performing monitoring on a dynamic cycle; Intelligent early warning module, which is used to receive monitoring data from fixed edge nodes and mobile edge nodes in real time, build a fault prediction model based on support vector machines, and predict the fault probability value; The fault warning module is used to compare the fault probability value with the preset multi-level warning threshold value, determine the fault level, and output warning information.

2. The charging fault real-time diagnosis system based on edge computing according to claim 1 is characterized in that: The charging fault data includes voltage data, current data, battery data, and device status data; The voltage data includes: real-time voltage value, voltage fluctuation amplitude, and voltage harmonic content; The current data includes: real-time current value, current change rate, and current harmonic distortion rate; The battery data includes: battery surface temperature, battery temperature gradient, battery state of charge, and battery health status; The device status data includes: plug-in and plug-out times, contact resistance, working mode, and fault indicator light status.

3. The real-time charging fault diagnosis system based on edge computing according to claim 1 is characterized in that: The data preprocessing module includes: Noise filtering unit, used to eliminate power frequency interference and high-frequency noise using adaptive LMS algorithm; Normalization processing unit, used to normalize the voltage and current signals to the [0,1] interval; Outlier correction unit, used to identify and replace abnormal data points using the 3σ criterion; A feature extraction unit is used to extract time domain features and frequency domain features from the charging fault data.

4. The real-time charging fault diagnosis system based on edge computing according to claim 1 is characterized in that: The risk grading module constructs a fault risk assessment model including: Determine the network topology, reserve pool parameter configuration, and calculate the reserve pool status; Perform forward propagation on the training sequence according to the reservoir state, record the reservoir state x(n) of all time steps, and output the fault risk level label y(n); Construct the reservoir state matrix X and output matrix Y according to the reservoir state x(n) of all time steps, and calculate the optimal output weight W using Tikhonov regularization out ; The optimal output weight W out It was integrated into the fault risk assessment model, and the generalization ability of the model was evaluated using five-fold cross validation to obtain the fault risk assessment model.

5. The real-time charging fault diagnosis system based on edge computing according to claim 1 is characterized in that: The risk grading module uses the pollen propagation algorithm to optimize the model hyperparameters including: Determine the echo state network hyperparameters that need to be optimized and set the value range of each parameter; Randomly generate M pollen individuals, each of which is defined as a combination of a set of ESN hyperparameters to form the initial population; Using the cross-validation accuracy as the evaluation indicator, the ESN model corresponding to each pollen individual was trained and validated, and its fitness value was calculated; Iterate for each pollen individual; Compare the fitness values ​​of all pollen individuals and retain the pollen individual with the highest fitness value as the optimal solution; Determine whether the maximum number of iterations has been reached or the fitness value has converged. If so, stop the optimization process and output the global optimal solution as the final hyperparameter. Otherwise, continue iterating.

6. The real-time charging fault diagnosis system based on edge computing according to claim 5 is characterized in that: In the risk grading module, the iterative process for each pollen individual includes: For each pollen individual, generate a random number and determine whether the random number is less than the probability p. If so, perform global pollination and update the solution according to the Levy flight. Otherwise, perform local pollination and update between adjacent solutions. Determine whether the updated solution exceeds the parameter value range, and if so, project it back into the boundary.

7. The real-time charging fault diagnosis system based on edge computing according to claim 6 is characterized in that: The formula of the Lévy flight update solution is expressed as: Where L is the step size of the Levy distribution, is the position of the global optimal solution in the entire population at the tth iteration, is the position of the i-th solution at the t-th iteration, represents the updated position of the i-th solution at the t+1-th iteration.

8. The real-time charging fault diagnosis system based on edge computing according to claim 1 is characterized in that: The dynamic monitoring module optimizes the layout position of fixed edge nodes including: Set the center coordinate set of the high-risk area to Q and the edge node monitoring radius to R; Initialize the empty set D and uncovered area set V; Select region q from V i , calculate q i The maximum straight-line distance to all other regions in V; Judge d imax Is it less than 2R? If so, set q i Add D and remove all distances q from V i Area smaller than R; Repeatedly optimize the fixed edge node layout until V is empty, and the output set D is the minimum number of edge nodes required.

9. The real-time charging fault diagnosis system based on edge computing according to claim 1 is characterized in that: In the dynamic monitoring module, the path planning of the mobile monitoring area includes: The coordinates of n sampling points that collect the boundary of the mobile monitoring area are set as the sampling point set S; The regional boundary curve is fitted using a cubic spline function based on the distribution density of the medium and low risk sub-regions and the sampling point set S; Determine the minimum rectangular range that can completely cover the area based on the boundary curve, generate uniform grid points within the rectangular range with the effective monitoring width of the mobile edge node as the interval, and select points that are located within the boundary curve and correspond to the centers of medium and low-risk sub-areas to form a set of must-pass points; A genetic algorithm is used to optimize the path of the necessary points, with the goal of minimizing the total length of the path and covering all low- and medium-risk sub-areas. The path sequence is iteratively optimized, and a dynamic weighting factor is introduced to give higher weights to low- and medium-risk sub-areas that have recently experienced minor anomalies, so that they are covered first in the inspection route. Adjust the inspection cycle according to the season and time period to ensure that mobile monitoring can provide comprehensive coverage and focus on potential risk points.

10. The real-time charging fault diagnosis system based on edge computing according to claim 3 is characterized in that: The fault prediction model construction of the intelligent early warning module includes: The extracted time domain features and frequency domain features are formed into a feature vector set; Dividing the feature vector set into a training set and a validation set, constructing a support vector machine model using the training set, selecting a radial basis kernel function as the kernel function, and optimizing the penalty coefficient C and kernel function parameter γ of the support vector machine model by a grid search method; Use the validation set to evaluate the performance of the optimized support vector machine model; The support vector machine model that passes the evaluation is the fault prediction model. The feature vector of the real-time monitoring data is input to obtain the fault probability value.

Images