Safety management monitoring system and method for charging facility

Abstract

The invention discloses a safety management monitoring system and method for a charging facility, and belongs to the technical field of charging piles. The system comprises an operation monitoring unit, an edge computing unit and a cloud platform. The operation monitoring unit obtains operation state data of each charging facility in the charging cluster area. The edge calculation unit is connected with the operation detection unit, and performs fault risk prediction and residual life prediction on each charging facility according to the operation state data to obtain a corresponding fault risk prediction result and a residual life prediction result; according to the fault risk prediction result, the residual life prediction result and a preset early warning judgment rule, whether early warning information is generated is judged, and if yes, the early warning information is sent to a cloud platform. And the cloud platform is connected with the edge computing unit, and completes safety management monitoring of each charging facility in the charging cluster area according to the early warning information. The safety management level of the charging facility can be remarkably improved, the fault risk is reduced, and the overall operation efficiency is improved.

Description

Technical Field

[0001] The present invention relates to a safety management and monitoring system and method for charging facilities, belonging to the technical field of charging piles. Background Art

[0002] As the global energy structure undergoes a profound transformation toward a low-carbon economy and environmental awareness becomes widespread, electric vehicles, with their outstanding characteristics as clean energy transportation, such as zero emissions and low noise, are experiencing unprecedented development opportunities. As a key player in the global electric vehicle market, my country has effectively stimulated the market through the coordinated efforts of policy guidance and market mechanisms, driving the continued steady growth of electric vehicle ownership. As a key supporting infrastructure for the popularization of electric vehicles, the number and scale of charging stations have also expanded significantly.

[0003] However, with the rapid increase in the number of charging pile facilities, the limitations of the traditional model of relying on manual inspections and record-keeping to collect facility data and monitor operational status are becoming increasingly apparent. Manual inspections are not only labor-intensive but also require long inspection cycles, making it difficult to promptly detect and address equipment failures, which in turn affects the normal utilization of charging piles. Manual record-keeping methods are prone to data omissions or errors, resulting in delayed information updates and a lack of comprehensive and accurate reflection of the charging pile's operational status and historical data. Furthermore, traditional methods are unable to meet the complex data collection and status monitoring requirements of large-scale charging pile facilities, with significant shortcomings in multiple aspects such as data collection, status monitoring, information integration, process processing, and inventory management.

[0004] Although existing charging pile management systems have achieved a certain degree of automation, they still suffer from numerous deficiencies in practical applications. Specifically, the existing system lacks accurate and effective analytical tools for real-time monitoring, fault warning, and anomaly diagnosis of charging piles, which, to a certain extent, affects the safety and stability of charging equipment. Its relatively simple functions for inspection plan formulation, task execution, and monitoring make it difficult to achieve full process automation, resulting in low overall operational efficiency. Fragmented archival information and the lack of a unified management platform restrict the efficiency of information updates and inquiries. Functions such as contract management, arrival planning, inspection and testing, and return and exchange processing are fragmented and lack an integrated management system, making business processes cumbersome and error-prone. Spare parts and equipment inventories still rely on manual management, lacking intelligent inventory monitoring and early warning mechanisms, which impacts the timeliness of equipment maintenance and troubleshooting.

[0005] Given this, existing technologies are unable to meet the needs of efficient data collection and status monitoring for large-scale charging facilities. Therefore, it is particularly urgent to develop an integrated and intelligent safety management and monitoring system for charging facilities. Summary of the Invention

[0006] The purpose of the present invention is to provide a safety management and monitoring system and method for charging facilities, aiming to achieve comprehensive monitoring and efficient control of charging facilities, and provide strong support for the sustainable development of the electric vehicle industry.

[0007] In order to solve the above technical problems, the present invention is implemented by adopting the following technical solutions: In one aspect, the present invention provides a safety management and monitoring system for charging facilities, comprising: an operation monitoring unit, an edge computing unit, and a cloud platform; Operation monitoring unit, used to obtain the operation status data of each charging facility in the charging cluster area; The edge computing unit is in communication with the operation detection unit and is used to predict the failure risk and remaining life of each charging facility based on the received operation status data, obtain corresponding failure risk prediction results and remaining life prediction results, and then determine whether to generate warning information based on the failure risk prediction results, remaining life prediction results and preset warning judgment rules. If so, the warning information is sent to the cloud platform; The cloud platform is connected to the edge computing unit for communication and is used to complete the safety management and monitoring of each charging facility in the charging cluster area based on the received warning information.

[0008] Optionally, the operating status data includes: cable temperature monitoring data, transformer monitoring data and charging waveform monitoring data; The cable temperature monitoring data is collected by distributed optical fiber temperature sensors; the distributed optical fiber temperature sensors are evenly distributed at preset intervals at the connectors of the charging pile cables and at the current load concentration section of the charging pile cables, and are used to dynamically monitor the temperature of the cables in real time, thereby generating cable temperature monitoring data; The transformer monitoring data is collected by a surface acoustic wave sensor embedded in the transformer insulating oil; the surface acoustic wave sensor dynamically monitors the dielectric properties of the insulating oil in real time by detecting the offset of the surface acoustic wave resonant frequency, thereby generating the transformer monitoring data; The charging waveform monitoring data includes: voltage distortion rate and current harmonic content; the charging waveform monitoring data is collected by a high-frequency harmonic acquisition unit set in the charging circuit of the charging pile; the high-frequency harmonic acquisition unit performs real-time dynamic detection of the current waveform data and voltage waveform data in the charging circuit at a sampling frequency of ≥100Hz, and then generates the voltage distortion rate and current harmonic content respectively.

[0009] Optionally, the remaining life prediction result is calculated using the following formula: ; Where, Remaining life prediction result, represents the failure health threshold, Indicates the current health value, represents the degradation rate predicted by the LSTM network, represents the standard deviation of random fluctuations introduced by the Wiener process, represents the standard normal noise introduced by the Wiener process.

[0010] Optionally, the failure risk of each charging facility is predicted using an improved Transformer model; the improved Transformer model adds a weighting mechanism to the multi-head attention mechanism to assign different specific weights to the relevant data at a specific moment; wherein the relevant data at a specific moment includes at least: the time point data corresponding to before and after the historical failure; the improved Transformer model also adds a residual connection in the encoder to alleviate the gradient vanishing of the deep network during the training process.

[0011] Optionally, the improved Transformer model is used to predict the failure risk of each charging facility, including: Acquire cable temperature monitoring data and transformer monitoring data; use a convolutional layer to extract local temperature change features from the cable temperature monitoring data, and use a Fourier transform method to extract frequency domain features from the transformer monitoring data; fusing the local temperature change feature and the frequency domain feature to obtain a comprehensive feature vector; Based on the weighting mechanism, weighting the comprehensive feature vector with relevant data at a specific moment to obtain a weight vector; The key vector and the value vector are multiplied by each element in the weight vector according to the time point to obtain a first eigenvector and a second eigenvector, and then a prediction result of the fault risk is obtained based on the first eigenvector and the second eigenvector.

[0012] Optionally, judging whether to generate warning information based on the failure risk prediction result, the remaining service life prediction result and a preset warning determination rule includes: If the remaining life prediction value corresponding to the remaining life prediction result of a charging facility is less than or equal to 7 days, a first-level remaining life warning information is generated; if the probability prediction value of a certain type of failure risk corresponding to the failure risk prediction result of a charging facility is greater than or equal to 85%, a first-level failure risk warning information is generated; If the remaining life prediction value corresponding to the remaining life prediction result of a charging facility is greater than 7 days and less than or equal to 30 days, a second-level remaining life warning information is generated; if the probability prediction value of a certain type of failure risk corresponding to the failure risk prediction result of a charging facility is greater than or equal to 60% and less than 85%, a second-level failure risk warning information is generated; If the remaining life prediction value corresponding to the remaining life prediction result of a charging facility is greater than 30 days but the weekly health decline rate is greater than 5%, a third-level remaining life warning information is generated; if the probability prediction value of a certain type of failure risk corresponding to the fault risk prediction result of a charging facility is greater than or equal to 30% and less than 60%, a third-level fault risk warning information is generated.

[0013] Optionally, the safety management and monitoring system for charging facilities further includes an operation and maintenance management unit communicatively connected to the cloud platform; The operation and maintenance management unit is configured to determine whether to generate an inspection route and a maintenance work order for the component to be replaced based on the remaining life warning information at each level, and to determine whether to generate an inspection route and a maintenance work order for the component to be repaired based on the failure risk warning information at each level; if so, the inspection route and the maintenance work order for the component to be replaced, and the inspection route and the maintenance work order for the component to be repaired are sent to an operation and maintenance terminal that is communicatively connected to the cloud platform; When the maintenance work order is completed, the corresponding operation and maintenance operation record will be synchronized to the operation monitoring unit for dynamic adjustment of the operation monitoring strategy.

[0014] Optionally, after the maintenance work order is completed, the operation and maintenance operation record will also be synchronized to the file management unit that is in communication with the cloud platform to form a maintenance history; The file management unit is used to uniformly manage the file information of each charging facility in the charging cluster area, and to store the production batches, inspection reports and operation and maintenance operation records of each charging facility in the charging cluster area based on blockchain technology, as well as the operation and maintenance operation records sent back by the operation and maintenance management unit.

[0015] Optionally, the safety management and monitoring system for charging facilities also includes a business management unit that is communicatively connected to the cloud platform; the business management unit utilizes an intelligent workflow engine to automatically process contract approval, arrival acceptance, and return and exchange applications, and during the contract creation and arrival plan formulation process, calls the archive information of each charging facility in the archive management unit and the inventory data of the warehousing management unit to achieve coordinated optimization of procurement and inventory.

[0016] Optionally, the warehouse management unit is also communicatively connected to the cloud platform; when the operation and maintenance personnel of the operation and maintenance terminal complete the replacement of relevant components in the charging facility, the warehouse management unit reversely drives the business management unit to generate a purchase order based on the updated spare parts requirements.

[0017] On the other hand, the present invention further provides a monitoring method applicable to the safety management and monitoring system for charging facilities as described in the first aspect, comprising: Obtain the operating status data of each charging facility in the charging cluster area; Based on the received operating status data, the fault risk and remaining life of each charging facility are predicted respectively to obtain corresponding fault risk prediction results and remaining life prediction results; Based on the fault risk prediction results, remaining life prediction results and preset warning judgment rules, it is determined whether to generate warning information, and then the safety management and monitoring of each charging facility in the charging cluster area is completed based on the judgment results.

[0018] Compared with the prior art, the present invention has the following beneficial effects: The present invention builds an efficient collaborative mechanism from the end side to the cloud side, forming a complete technical chain of "monitoring-analysis-management-optimization": multi-source sensors are used to collect the electrical, thermal and mechanical status data of the charging piles in real time on the end side. After the edge computing unit pre-processes the data, it relies on the intelligent model to complete the remaining life prediction, accurate identification of fault risks and graded warning. The cloud platform is responsible for data storage, in-depth analysis and cross-unit linkage, integrating file management, warehouse scheduling and other functions to achieve full-cycle management of equipment status. Compared with traditional solutions, this system breaks through the technical bottlenecks of poor real-time performance, high false alarm rate of faults, and delayed operation and maintenance response of traditional monitoring without increasing hardware complexity through multi-source sensor fusion and intelligent algorithm optimization. It allows operation and maintenance personnel to intuitively grasp the equipment status, complete abnormal diagnosis, trace maintenance records and dispatch intelligent storage resources through the GIS visualization interface. Efficient management of the entire process can be achieved without relying on dedicated clients, providing full-chain technical support from data perception to decision-making for the safe and stable operation of the smart charging network. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 FIG2 is a schematic diagram showing a framework of an embodiment of a safety management and monitoring system for charging facilities according to the present invention; Figure 2 Schematic diagram of the process of predicting the remaining life in the safety management and monitoring system for charging facilities of the present invention; Figure 3 Shown is a flow chart of each type of fault risk prediction in the safety management and monitoring system for charging facilities of the present invention. DETAILED DESCRIPTION

[0020] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations on the technical solution of the present invention. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0021] The term "and / or" simply describes a relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can mean: A exists alone, A and B exist simultaneously, or B exists alone. Additionally, the character " / " generally indicates an "or" relationship between the related objects.

[0022] Example 1 This embodiment introduces a safety management and monitoring system for charging facilities, which includes: an operation monitoring unit, an edge computing unit and a cloud platform.

[0023] The operation monitoring unit is used to obtain the operation status data of each charging facility in the charging cluster area.

[0024] The edge computing unit is communicatively connected to the operation detection unit and is used to predict the failure risk and remaining life of each charging facility based on the received operation status data, obtain corresponding failure risk prediction results and remaining life prediction results, and then determine whether to generate early warning information based on the failure risk prediction results, remaining life prediction results and preset early warning judgment rules. If so, the early warning information is sent to the cloud platform.

[0025] The cloud platform is connected to the edge computing unit for communication and is used to complete the safety management and monitoring of each charging facility in the charging cluster area based on the received warning information.

[0026] During the actual application of the charging facility safety management and monitoring system of this embodiment, the operation monitoring unit first comprehensively acquires the operating status data of each charging facility in the charging cluster area. This data includes but is not limited to cable temperature monitoring data, transformer monitoring data, and charging waveform monitoring data. These data are collected in real time and dynamically through distributed optical fiber temperature sensors, surface acoustic wave sensors embedded in the transformer insulating oil, and high-frequency harmonic acquisition units set in the charging circuit of the charging pile, thereby ensuring the accuracy and timeliness of the data. Subsequently, this operating status data is transmitted to the edge computing unit that is communicatively connected to the operation monitoring unit. The edge computing unit uses an improved Transformer model to predict the failure risk of each charging facility and uses a specific formula to calculate the remaining life prediction result of each charging facility. After obtaining the failure risk prediction result and the remaining life prediction result, the edge computing unit further determines whether it is necessary to generate an early warning information based on the preset early warning judgment rules. Once it is determined that an early warning information needs to be generated, the edge computing unit will immediately send the early warning information to the cloud platform with which it is in communication. Finally, after receiving the early warning information, the cloud platform completes the safety management and monitoring of each charging facility in the charging cluster area based on this information, including but not limited to the processing of early warning information, the allocation of operation and maintenance tasks, and the continuous tracking of the operating status of charging facilities, so as to ensure the safe and stable operation of the charging facilities and provide strong guarantees for the charging services of electric vehicles.

[0027] Example 2 like Figure 1 As shown, this embodiment utilizes the "end-side information acquisition - edge computing processing - cloud-based collaborative management" process as its core architecture, aiming to achieve real-time monitoring of charging pile operating status and refined management throughout its entire lifecycle. The system's structure fully considers the functional requirements and coordination of each link. Its specific structure and functional implementation process are described below.

[0028] 1. Acquisition of Multi-Source Sensor Data on the Device Side Cable temperature monitoring data: DTS (distributed fiber temperature sensing) sensors are evenly spaced 1 meter apart at the connectors of the charging pile cables (such as where the cable connects to the internal wiring of the charging pile and where the cable connects to the charging gun), as well as in concentrated load sections (typically areas with high cable current density). These sensors operate based on the principle of fiber backscattering, continuously collecting the temperature distribution along the entire cable with a spatial resolution of 1 meter and an accuracy of ±0.5°C. In actual operation, when poor cable contact causes increased local resistance or an abnormal increase in current due to overload, the temperature in these areas will rise abnormally. The temperature data collected by the sensors will serve as a key variable for determining the operating status of the cable and will be subsequently transmitted to edge computing nodes for fault risk identification and remaining life prediction.

[0029] Transformer monitoring data: SAW (surface acoustic wave) sensors are embedded in the transformer's insulating oil. The SAW sensors dynamically monitor the dielectric properties of the insulating oil by detecting the offset in the surface acoustic wave resonant frequency. When the insulating oil ages (such as through oxidation and decomposition caused by long-term operation) or becomes contaminated (such as through the incorporation of moisture or impurities), its dielectric properties change, causing the SAW sensor's frequency response to change accordingly. The collected dielectric property data accurately reflects the insulation condition within the transformer and is also transmitted to edge computing nodes to contribute to the overall health assessment of the equipment. Charging waveform monitoring data: A high-frequency harmonic acquisition unit is installed in the charging circuit of the charging pile. This unit collects the current and voltage waveforms during the charging process at a sampling frequency of ≥100Hz. By analyzing the collected waveform data, high-frequency distortion characteristics such as voltage distortion rate and current harmonic content can be obtained. The voltage distortion rate reflects the degree to which the voltage waveform deviates from the ideal sine wave, while the current harmonic content reflects the magnitude of the various harmonic components in the current. This high-frequency distortion characteristic data will provide a basis for subsequent abnormal discharge and harmonic pollution analysis, and is also an important basis for fault risk identification and remaining life prediction. It will be transmitted to the edge computing node for further processing.

[0030] In addition, this embodiment also monitors the equipment's environment and auxiliary equipment. Specifically, temperature, humidity, smoke, and hazardous gas sensors are deployed to provide real-time monitoring of charging station environmental parameters. When humidity and temperature exceed warning values, the ventilation system is automatically activated and the equipment's moisture-proof status is marked. The operating status of auxiliary facilities such as charging station awnings and fire sprinklers is monitored. In the event of anomalies, fault locations are mapped and pushed to the operation and maintenance terminal via a GIS map.

[0031] 2. Edge Data Processing and Intelligent Analysis After the cable temperature monitoring data, transformer monitoring data, and charging waveform monitoring data collected on the end side are transmitted to the edge computing node, corresponding data preprocessing operations are required, as described below.

[0032] Denoising: A sliding average filter is used to denoise cable temperature monitoring data, transformer monitoring data, and charging waveform monitoring data. Taking cable temperature monitoring data as an example, a sliding window size of five sampling points is set. The average temperature data within the window is taken as the new temperature value at the center of the window. This process is repeated for the entire temperature data sequence, eliminating random noise caused by factors such as environmental interference. Normalization: Normalize all collected raw data and map them to the [0, 1] interval. The specific formula is: ; Where, represents the normalized data, Represents the original data, and The normalization process is to unify the dimensions of different types of data to facilitate subsequent model processing.

[0033] Outlier and missing value processing: Outliers are identified by setting thresholds. For example, for cable temperature data, if the temperature at a given moment exceeds the normal operating temperature range (e.g., exceeding 80°C; this threshold can be adjusted based on actual operating experience), it is considered an outlier and removed. Missing values are filled using linear interpolation, using linear calculations based on the data points before and after the missing value to determine the fill value, ensuring data integrity and continuity.

[0034] Furthermore, this embodiment will sequentially perform remaining life prediction and failure risk prediction for each charging facility in the charging cluster based on the preprocessed data. The specific implementation process is as follows.

[0035] The remaining life prediction process uses a model that combines LSTM (Long Short-Term Memory) and the Wiener process. Specifically, the LSTM network learns the time series characteristics of historical data through multiple hidden layers, effectively capturing the temporal changes in temperature and load rate and exploring long-term dependencies in the data. The Wiener process is used to describe the random degradation trend of the equipment's health status. By combining the patterns learned by the LSTM network and the random degradation characteristics described by the Wiener process, a remaining life prediction equation is established, such as Figure 2 shown.

[0036] In actual application, 3 months of historical data are divided into 12 segments by week. Each segment has an input sequence length of 168 hours (7 days) and a time step of 1 hour. A 3-layer LSTM network is used to extract temporal dependencies and learn degradation patterns. The fully connected layer outputs a deterministic degradation rate. Then, the Wiener process introduces random terms to generate health state increments Finally, the remaining useful life (RUL) is calculated based on the health gap and the total degradation rate: ; Where, Remaining life prediction result, represents the failure health threshold, Indicates the current health value, represents the degradation rate predicted by the LSTM network, represents the standard deviation of random fluctuations introduced by the Wiener process, represents the standard normal noise introduced by the Wiener process.

[0037] In terms of model selection for the fault risk prediction process, the traditional Transformer model is improved by introducing a weighted mechanism into the multi-head attention mechanism. Different weights are assigned to data at different time points based on the importance of historical data, allowing the model to pay more attention to recent data and data that has a greater impact on fault judgment. At the same time, residual connections are added to the encoder part of the model to alleviate the gradient vanishing problem during deep network training and improve the training efficiency and performance of the model. The overall processing flow is as follows: Figure 3 shown.

[0038] Introduction of weighting mechanism in multi-head attention mechanism: In the traditional Transformer model, the multi-head attention mechanism treats data at each time point equally. However, in the fault risk identification scenario, the importance of data at different time points for fault judgment varies. Recent data can often reflect the current operating status of the equipment more promptly, while data at certain specific moments may have a key impact on fault judgment. Therefore, we introduce a weighting mechanism. Specifically, by analyzing historical data, methods such as time decay functions are used to measure the importance of data at each time point. For example, the closer the data is to the time, the higher the weight is given; for data at key time points before and after historical faults, the weight is also increased accordingly. When calculating the attention score, it is no longer a simple dot product operation, but the data vector of each time point is multiplied by the corresponding weight before the operation is performed. Assume that the input time series data is , the corresponding weight vector is , then when calculating the attention score, for the query vector , key vector Sum value vector , first and Multiply the weight vector elements at each time point to get the first eigenvector and the second eigenvector Then, according to the traditional multi-head attention mechanism calculation method, The weighted attention output enables the model to focus more on important data and improve its ability to capture fault characteristics.

[0039] Adding residual connections to the encoder: As the depth of the Transformer model increases, the vanishing gradient problem may cause training difficulties and affect model performance. To solve this problem, residual connections are introduced in the encoder part of the model. For each layer in the encoder, assuming the input is , the output after the attention mechanism and feedforward neural network operations is , then the residual connection makes the final output of this layer become . In this way, during the back propagation process, the gradient can be directly passed through This path of transmission avoids the vanishing gradient caused by multi-layer nonlinear transformations, enabling more efficient training of deep networks. Furthermore, residual connections help the model learn identity mappings, preventing overfitting to a certain extent, improving training efficiency and generalization capabilities, and ultimately enhancing the accuracy of fault risk identification.

[0040] Specifically, the preprocessed data is input into the improved Transformer model. In the input layer of the model, feature extraction is performed for different types of data. For cable temperature monitoring data, the convolution layer is used to extract local temperature change features; for transformer monitoring data, frequency domain features are extracted through Fourier transform and other methods. The features of the extracted different types of data are then fused to form a comprehensive feature vector as the input of the multi-head attention mechanism. The fused feature vector enters the multi-head attention mechanism. Under the action of the weighting mechanism, the model focuses on the data features of important time points and captures the correlation between different features. The data processed by the multi-head attention mechanism then enters the feedforward neural network for further nonlinear transformation and feature extraction. Finally, the model passes through a fully connected layer and The activation function outputs a probability prediction value of the failure risk. For example, the output result is a probability vector ,in, represents the number of different fault types, Indicates that a charging facility has The probability values for various types of failures are presented. Failure risk types include, but are not limited to, poor contact, harmonic overload, and insulation aging. In actual applications, these probability values can be used to determine the current failure risk status of the device.

[0041] Furthermore, it is possible to determine whether to generate warning information based on the failure risk prediction results, the remaining life prediction results, and the preset warning judgment rules, including the following situations: If the remaining life prediction result of a charging facility corresponds to a predicted value of less than or equal to 7 days, a Level 1 remaining life warning message is generated. If the probability prediction value of a certain type of failure risk corresponding to the failure risk prediction result of a charging facility is greater than or equal to 85%, a Level 1 failure risk warning message is generated.

[0042] If the remaining life prediction result of a charging facility corresponds to a predicted value of more than 7 days and less than or equal to 30 days, a Level 2 remaining life warning message is generated. If the probability prediction value of a certain type of failure risk corresponding to the failure risk prediction result of a charging facility is greater than or equal to 60% and less than 85%, a Level 2 failure risk warning message is generated.

[0043] If the remaining life prediction result for a charging facility is greater than 30 days but the weekly health decline rate is greater than 5%, a Level 3 remaining life warning message is generated. If the probability prediction value for a certain type of failure risk corresponding to the failure risk prediction result for a charging facility is greater than or equal to 30% and less than 60%, a Level 3 failure risk warning message is generated.

[0044] Among them, various types of fault risk warning information include at least: fault code, location, and links to historical similar cases. The above historical similar case links are from the knowledge graph of the archive management unit to shorten the fault location time.

[0045] 3. Cloud Data Integration and Multi-Unit Collaborative Management refer to Figure 1 ,After receiving the edge computing results, the cloud platform performs the following operations: Storing and analyzing global data: The operating status data, fault risk status, remaining life status, and other data of all charging facilities in the area are stored in a distributed database to support historical data backtracking and trend analysis.

[0046] Equipment health management: Based on the GIS visualization interface, the equipment health status is displayed at the "region-charging station-charging pile" level. Operation and maintenance personnel can click on a single device to view real-time parameters such as voltage, current, and power.

[0047] Identify fault clusters through spatiotemporal correlation analysis: Using a spatiotemporal correlation analysis algorithm, if similar faults (such as harmonic pollution) occur at multiple charging stations in a certain area within 24 hours, the spatiotemporal distribution characteristics of the faults are automatically analyzed to determine whether common problems are caused by grid fluctuations or environmental factors, triggering an emergency work order that is pushed to operation and maintenance personnel.

[0048] The cloud platform is also communicatively connected to an operations management unit, a file management unit, a business management unit, and a storage management unit. Through the integration of these units, this embodiment enables comprehensive monitoring and management of charging facilities, significantly improving management efficiency and reducing operations and maintenance costs, thus meeting the requirements for efficient management of large-scale charging pile facilities. The functional implementation of each unit is described in detail below.

[0049] Operation and maintenance management unit: includes inspection plan query, inspection plan execution status, inspection plan details, inspection task execution and new inspection plan functions. Users can use this unit to formulate inspection plans, monitor the execution of inspection tasks, and conduct inspections and maintenance management of charging pile facilities to ensure the normal operation of the equipment. In actual application, this unit determines whether to generate inspection routes and maintenance work orders for parts to be replaced based on the remaining life warning information at all levels, and determines whether to generate inspection routes and maintenance work orders for parts to be repaired based on the fault risk warning information at all levels; if so, the inspection routes and maintenance work orders for parts to be replaced and the inspection routes and maintenance work orders for parts to be repaired are sent to the operation and maintenance terminal that is connected to the cloud platform.

[0050] It's worth noting that the system also features intelligent inspection scheduling, integrating Geographic Information System (GIS) data with real-time fault analysis. GIS displays the distribution of charging stations in real time, correlating them with fault risk levels. The system analyzes fault distribution, inspector location, and skills, optimizes inspection routes and task allocation, and pushes tasks, including routes and fault prediction guidelines, to terminals, improving inspection efficiency. In other words, in this embodiment, inspection route generation requires integration with GIS maps, and the generated routes prioritize high-risk equipment. Regarding intelligent inspection scheduling, this unit also supports manual adjustments and task assignments, and real-time tracking of inspection progress, such as QR code sign-in and fault photo upload. Upon completion of a maintenance work order, the corresponding operation and maintenance records (parts replaced, time consumption, etc.) are synchronized to the operation monitoring unit for dynamic adjustment of the operation monitoring strategy. Furthermore, upon completion of a maintenance work order, these operation and maintenance records are synchronized to the archive management unit, which is connected to the cloud platform, to form a maintenance history.

[0051] Archives Management Unit: This integrates archive statistics and management functions, enabling multi-dimensional statistical analysis and visual display of archives of charging stations, charging piles, and other facilities to assist in decision-making. It also supports centralized archive management, including query, editing, and import operations, ensuring the accuracy and completeness of archive information and enabling digital management and control of the entire lifecycle of charging facilities. In actual application, this unit centrally manages the archive information of each charging facility within the charging cluster area, and uses blockchain technology to store production batches, test reports, and operation and maintenance records transmitted back by the operation and maintenance management unit for each charging facility within the charging cluster area, ensuring that the data cannot be tampered with.

[0052] Secondly, this unit also receives fault risk events and operation and maintenance operation records, and uses the knowledge graph to associate "equipment model-failure mode-maintenance plan" to achieve defect traceability. For example, if a batch of charging guns frequently causes insulation failures, the spare parts replacement records are automatically associated. Among them, the above-mentioned knowledge graph collects charging pile equipment information, fault records, maintenance records, and defines entity relationships. When a device fails, the system searches the graph. If it is found that the historical fault correlation of the same type of equipment is high, it automatically associates the relevant defect entities and generates inspection and processing suggestions. The graph is updated through the graph neural network algorithm to improve the accuracy of defect traceability. In addition, this unit also uses visualization to display indicators such as equipment failure rate trends, operation and maintenance efficiency (such as average fault repair time), inventory turnover rate, etc., to assist management in formulating equipment update plans and resource allocation strategies.

[0053] The Business Management Unit utilizes an intelligent workflow engine to automate contract approvals, arrival inspections, and return and exchange requests. For example, upon arrival, the unit automatically links contract terms with inspection standards by scanning the QR code of the device, generating an acceptance report. Furthermore, during contract creation and delivery planning, the unit utilizes the archival information of each charging facility and the inventory data from the Warehousing Management Unit to achieve coordinated optimization of procurement and inventory. Furthermore, this unit integrates equipment operation data (from the Operation Monitoring Unit) to intelligently recommend technical improvement solutions (e.g., automatically recommending capacity expansion for high-load charging stations), reducing manual decision-making costs.

[0054] The Warehouse Management Unit includes warehouse management, incoming and outgoing inventory management, and inventory detail display. Users can use this unit to manage the spare parts and equipment inventory of charging pile facilities, supporting operations such as incoming and outgoing inventory, and inventory queries, ensuring optimal inventory allocation and management. In actual use, when maintenance personnel at the operation and maintenance terminal complete replacement of relevant components in the charging facility, the Warehouse Management Unit drives the Business Management Unit to generate purchase orders based on the updated spare parts demand. Furthermore, this unit uses the LSTM-DRL intelligent warehouse scheduling algorithm based on edge computing fault prediction results to optimize replenishment strategies. Specifically, the LSTM analyzes historical spare parts consumption data to predict demand, while the DRL optimizes replenishment strategies based on inventory turnover and out-of-stock rates. For example, when the failure risk of a certain charging pile model increases, the system automatically increases the corresponding spare part inventory. At the same time, an attention mechanism analyzes the correlation between region, model, and spare part demand to improve replenishment accuracy. For example, the safety stock can be dynamically adjusted based on the replacement frequency of a certain circuit breaker model over the past three months. When inventory falls below the safety threshold, a purchase requisition is automatically sent to the Business Management Unit. In addition, this unit also supports spare parts positioning (RFID tag management) and chain storage of inbound and outbound records. The flow of spare parts can be traced by scanning the code. For example, the "202308 batch contactor" is associated with 3 charging interruption faults.

[0055] In actual application, this embodiment collects charging pile operation data in real time through a multi-source sensor network on the end side, and combines it with an edge computing unit for rapid preprocessing and analysis, significantly improving the real-time performance of monitoring. At the same time, by utilizing intelligent algorithms such as the improved Transformer model and the LSTM-Wiener process model, accurate assessment of the electrical, thermal, and mechanical status of the charging pile is achieved, reducing the false alarm rate of faults. Secondly, this embodiment can also intelligently generate inspection plans and maintenance work orders based on the equipment health status and fault risk prediction results, and display the equipment status and inspection progress in real time through the GIS interface, allowing operation and maintenance personnel to quickly respond to and handle abnormal situations, significantly improving operation and maintenance efficiency. Thirdly, this embodiment also introduces a blockchain evidence storage system to store the entire life cycle archive of the charging pile facility on the chain, ensuring the data's non-tamperability and traceability. Combined with knowledge graph technology, the system can automatically associate equipment fault records, achieve accurate defect traceability analysis, and provide strong support for operation and maintenance decision-making. In addition, this embodiment also integrates an intelligent workflow engine to realize the automated approval and task flow of business processes such as contract management, arrival planning, identification and testing, and return and exchange management, reducing manual intervention and improving approval efficiency. At the same time, the system can also automatically optimize equipment technical transformation and maintenance plans based on equipment operating status and historical data, ensuring an efficient closed-loop business process. In addition, this embodiment accurately predicts inventory demand trends and optimizes inventory scheduling strategies by building an intelligent warehouse management system and combining LSTM with deep reinforcement learning algorithms. The system can predict spare parts demand based on operation and maintenance work orders and monitoring data, and reversely drive the business management unit to generate purchase orders, reducing inventory backlogs and out-of-stock rates and improving inventory turnover.

[0056] In summary, this embodiment leverages core technologies such as multi-dimensional sensor networks, edge intelligent diagnosis, blockchain data management, intelligent workflow engines, and intelligent warehouse monitoring to build a comprehensive intelligent charging facility safety management and monitoring system. This system not only significantly improves equipment monitoring accuracy and optimizes inspection efficiency, but also strengthens data security, simplifies business processes, and enhances the intelligence of warehouse management, ultimately driving the upgrade of charging pile management systems towards efficiency, intelligence, and sustainability.

[0057] Example 3 Based on the same technical concept as in Example 1 or 2, this embodiment provides a monitoring method applicable to the safety management and monitoring system for charging facilities as described in Example 1 or 2, which includes the following steps: Obtain the operating status data of each charging facility in the charging cluster area; Based on the received operating status data, the fault risk and remaining life of each charging facility are predicted respectively to obtain corresponding fault risk prediction results and remaining life prediction results; Based on the fault risk prediction results, remaining life prediction results and preset warning judgment rules, it is determined whether to generate warning information, and then the safety management and monitoring of each charging facility in the charging cluster area is completed based on the judgment results.

[0058] During the actual application of the monitoring method of this embodiment, the operating status data of each charging facility in the charging cluster area is first obtained in real time through the operation monitoring unit. This data covers multi-dimensional information such as the electrical parameters, thermal status, and mechanical operation status of the charging pile. Subsequently, the collected data is transmitted to the edge computing unit, which uses the built-in intelligent algorithm model to conduct an in-depth analysis of the operating status data of each charging facility, respectively completing the prediction of failure risk and remaining life, and generating corresponding prediction results. Based on these prediction results and the pre-set warning judgment rules, the edge computing unit further determines whether the warning mechanism needs to be triggered. Once the judgment conditions are met, the warning information is generated and uploaded to the cloud platform. After receiving the early warning information, the cloud platform combines the data from the file management, warehouse scheduling and other units to implement comprehensive safety management and monitoring of each charging facility in the charging cluster area, including but not limited to troubleshooting, operation and maintenance scheduling, resource allocation and other operations to ensure the safe and stable operation of the charging facilities. At the same time, operation and maintenance personnel can use the GIS visualization interface provided by the cloud platform to intuitively view the equipment status, perform abnormal diagnosis, manage maintenance records and dispatch intelligent warehousing, realizing efficient operation and maintenance management of the entire process. The entire process forms a closed loop from data collection, analysis and early warning to operation and maintenance management, effectively solving the problems of insufficient real-time monitoring, high false alarm rate of faults and delayed operation and maintenance response in traditional charging pile monitoring, and providing a solid technical guarantee for the safe operation of the smart charging network.

[0059] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A safety management and monitoring system for charging facilities, characterized in that: include: Operation monitoring unit, edge computing unit and cloud platform; Operation monitoring unit, used to obtain the operation status data of each charging facility in the charging cluster area; The edge computing unit is in communication with the operation detection unit and is used to predict the failure risk and remaining life of each charging facility based on the received operation status data, obtain corresponding failure risk prediction results and remaining life prediction results, and then determine whether to generate warning information based on the failure risk prediction results, remaining life prediction results and preset warning judgment rules. If so, the warning information is sent to the cloud platform; The cloud platform is connected to the edge computing unit for communication and is used to complete the safety management and monitoring of each charging facility in the charging cluster area based on the received warning information.

2. The charging facility safety management and monitoring system according to claim 1, characterized in that: The operating status data includes: cable temperature monitoring data, transformer monitoring data and charging waveform monitoring data; The cable temperature monitoring data is collected by distributed optical fiber temperature sensors; the distributed optical fiber temperature sensors are evenly distributed at preset intervals at the connectors of the charging pile cables and at the current load concentration section of the charging pile cables, and are used to dynamically monitor the temperature of the cables in real time, thereby generating cable temperature monitoring data; The transformer monitoring data is collected by a surface acoustic wave sensor embedded in the transformer insulating oil; the surface acoustic wave sensor dynamically monitors the dielectric properties of the insulating oil in real time by detecting the offset of the surface acoustic wave resonant frequency, thereby generating the transformer monitoring data; The charging waveform monitoring data includes: voltage distortion rate and current harmonic content; the charging waveform monitoring data is collected by a high-frequency harmonic acquisition unit set in the charging circuit of the charging pile; the high-frequency harmonic acquisition unit performs real-time dynamic detection of the current waveform data and voltage waveform data in the charging circuit at a sampling frequency of ≥100Hz, and then generates the voltage distortion rate and current harmonic content respectively.

3. The safety management and monitoring system for charging facilities according to claim 2, characterized in that: The remaining life prediction result is calculated by the following formula: ； Where, Remaining life prediction result, represents the failure health threshold, Indicates the current health value, represents the degradation rate predicted by the LSTM network, represents the standard deviation of random fluctuations introduced by the Wiener process, represents the standard normal noise introduced by the Wiener process.

4. The charging facility safety management and monitoring system according to claim 3, characterized in that: The improved Transformer model predicts the failure risk of each charging facility. The improved Transformer model incorporates a weighting mechanism into the multi-head attention mechanism to assign specific weights to relevant data at specific moments. The relevant data at specific moments includes at least the time points before and after the historical failure. The improved Transformer model also adds residual connections to the encoder to mitigate gradient vanishing during deep network training. The improved Transformer model is used to predict the failure risk of each charging facility, including: Acquire cable temperature monitoring data and transformer monitoring data; use a convolutional layer to extract local temperature change features from the cable temperature monitoring data, and use a Fourier transform method to extract frequency domain features from the transformer monitoring data; fusing the local temperature change feature and the frequency domain feature to obtain a comprehensive feature vector; Based on the weighting mechanism, weighting the comprehensive feature vector with relevant data at a specific moment to obtain a weight vector; The key vector and the value vector are multiplied by each element in the weight vector according to the time point to obtain a first eigenvector and a second eigenvector, and then a prediction result of the fault risk is obtained based on the first eigenvector and the second eigenvector.

5. The charging facility safety management and monitoring system according to claim 4, characterized in that: The determination of whether to generate warning information based on the failure risk prediction result, the remaining service life prediction result and the preset warning judgment rules includes: If the remaining life prediction value corresponding to the remaining life prediction result of a charging facility is less than or equal to 7 days, a first-level remaining life warning information is generated; if the probability prediction value of a certain type of failure risk corresponding to the failure risk prediction result of a charging facility is greater than or equal to 85%, a first-level failure risk warning information is generated; If the remaining life prediction value corresponding to the remaining life prediction result of a charging facility is greater than 7 days and less than or equal to 30 days, a second-level remaining life warning information is generated; if the probability prediction value of a certain type of failure risk corresponding to the failure risk prediction result of a charging facility is greater than or equal to 60% and less than 85%, a second-level failure risk warning information is generated; If the remaining life prediction value corresponding to the remaining life prediction result of a charging facility is greater than 30 days but the weekly health decline rate is greater than 5%, a third-level remaining life warning information is generated; if the probability prediction value of a certain type of failure risk corresponding to the fault risk prediction result of a charging facility is greater than or equal to 30% and less than 60%, a third-level fault risk warning information is generated.

6. The charging facility safety management and monitoring system according to claim 5, characterized in that: It also includes an operation and maintenance management unit that is in communication with the cloud platform; The operation and maintenance management unit is used to determine whether to generate an inspection route and a maintenance work order for the component to be replaced based on the remaining life warning information at each level, and to determine whether to generate an inspection route and a maintenance work order for the component to be repaired based on the failure risk warning information at each level; If so, the inspection route and maintenance work order of the component to be replaced, and the inspection route and maintenance work order of the component to be repaired are sent to the operation and maintenance terminal that is communicatively connected to the cloud platform; When the maintenance work order is completed, the corresponding operation and maintenance operation record will be synchronized to the operation monitoring unit for dynamic adjustment of the operation monitoring strategy.

7. The charging facility safety management and monitoring system according to claim 6, characterized in that: When a maintenance work order is completed, the operation and maintenance operation record will also be synchronized to the file management unit that is connected to the cloud platform to form a maintenance history; The file management unit is used to uniformly manage the file information of each charging facility in the charging cluster area, and to store the production batches, inspection reports and operation and maintenance operation records of each charging facility in the charging cluster area based on blockchain technology, as well as the operation and maintenance operation records sent back by the operation and maintenance management unit.

8. The charging facility safety management and monitoring system according to claim 7, characterized in that: It also includes a business management unit that is communicatively connected to the cloud platform; the business management unit uses an intelligent workflow engine to automatically transfer contract approval, arrival acceptance and return and exchange applications, and during the contract creation and arrival plan formulation process, calls the archive information of each charging facility in the archive management unit and the inventory data of the warehousing management unit to achieve coordinated optimization of procurement and inventory.

9. The charging facility safety management and monitoring system according to claim 8, characterized in that: The warehouse management unit is also connected to the cloud platform for communication; when the operation and maintenance personnel of the operation and maintenance terminal complete the replacement of relevant components in the charging facility, the warehouse management unit reversely drives the business management unit to generate a purchase order based on the updated spare parts requirements.

10. The monitoring method of the charging facility safety management and monitoring system according to any one of claims 1 to 9, characterized in that: include: Obtain the operating status data of each charging facility in the charging cluster area; Based on the received operating status data, the fault risk and remaining life of each charging facility are predicted respectively to obtain corresponding fault risk prediction results and remaining life prediction results; Based on the fault risk prediction results, remaining life prediction results and preset warning judgment rules, it is determined whether to generate warning information, and then the safety management and monitoring of each charging facility in the charging cluster area is completed based on the judgment results.

Images