Cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile group

Abstract

This invention discloses a cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters, belonging to the field of charging pile cluster operation and maintenance management technology. The system includes a cloud-based overall control module, a distributed edge collaboration module, a terminal device perception and execution module, and a cross-edge node cluster linkage sub-module. This invention addresses the pain points of existing technologies, such as lack of cloud-edge collaboration, network outage paralysis, algorithm rigidity, lack of resource coordination, and susceptibility to power grid imbalance, making it difficult to meet intelligent operation and maintenance needs. This invention combines the global decision-making advantages of the cloud-based overall control module with the real-time response capabilities of the edge, achieving optimal computing power configuration. Even after a network outage, edge nodes can independently complete all operation and maintenance management tasks, avoiding station shutdowns caused by single module failures. This ensures the continuous and stable operation of large-scale charging pile clusters, significantly improving operational reliability and achieving load sharing, fault redundancy switching, orderly charging, and power grid peak shaving and valley filling.

Description

Technical Field

[0001] This invention relates to the field of charging pile group operation and maintenance management technology, specifically to a cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile groups. Background Technology

[0002] With the explosive growth of the new energy vehicle industry, electric vehicle charging piles, as a core new infrastructure, have seen exponential expansion in construction scale. The layout pattern is gradually shifting from a decentralized deployment of single sites to a cross-regional, large-scale, high-density cluster model. Traditional operation and maintenance management systems mostly adopt an independent management and control model for single sites, which only has local equipment monitoring and basic fault alarm functions, resulting in significant industry pain points and technical deficiencies.

[0003] These systems typically employ centralized control via a pure cloud-based unified management module or operate in isolated single-edge environments, lacking a cloud-edge collaboration mechanism. The massive data uploads result in high bandwidth consumption and significant transmission latency, failing to meet the demands for high real-time management. They are also highly susceptible to paralysis after edge network outages, leading to poor service stability. Furthermore, their static and fixed algorithm models cannot rely on iterative processing based on full-domain data, resulting in low prediction accuracy and a passive response to faults, hindering proactive preventative maintenance. Simultaneously, the lack of global coordination of maintenance resources, the absence of cross-site load sharing and fault redundancy switching mechanisms, and high maintenance costs, low handling efficiency, and potential grid load imbalance make them unsuitable for the efficient and intelligent maintenance requirements of large-scale charging pile clusters. Therefore, they do not meet existing needs. To address this, we propose a cloud-edge collaborative intelligent maintenance management system for large-scale charging pile clusters. Summary of the Invention

[0004] The purpose of this invention is to provide a cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters. By constructing a cloud-edge collaborative system with global coordination and regional autonomy, it combines the global decision-making advantages of the cloud-based overall management and control module with the real-time response capabilities of the edge. It can complete high-real-time fault handling and local control in milliseconds, achieve optimal computing power configuration, and independently complete all operation and maintenance management tasks after the edge node is disconnected from the network. The dual redundancy design of hardware and cloud edge avoids the failure of a single module from causing the station to shut down, ensuring the continuous and stable operation of large-scale charging pile clusters, significantly improving operational reliability, realizing load sharing, fault redundancy switching, orderly charging and grid peak shaving and valley filling, balancing the grid load, and solving the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters, comprising:

[0006] The cloud-based overall management and control module is used to realize global computing power scheduling, full-scale big data analysis, intelligent algorithm iteration, global operation and maintenance strategy generation, cross-site resource coordination, and full life cycle asset management;

[0007] The distributed edge collaboration module is used to realize local real-time control, millisecond-level fault response, local computing power offloading, multi-source data preprocessing and caching functions;

[0008] The terminal device sensing and execution module is used to collect charging pile operating parameters, environmental data, charging status and fault information in real time, receive control commands issued by the distributed edge collaboration module and execute them accurately, and complete basic execution actions such as charging start and stop, power adjustment, fault reporting and status feedback.

[0009] The cross-edge node cluster linkage submodule is used to collect charging pile operating parameters, environmental data, charging status and fault information in real time, and to perform charging start and stop, power adjustment, fault reporting and status feedback.

[0010] Preferably, the cloud-based overall management and control module specifically includes:

[0011] The global computing power scheduling unit is used to monitor the cloud-based overall management and control module and the edge computing power load in real time, establish hierarchical scheduling rules, and dynamically allocate computing power resources;

[0012] The big data analytics unit is used to collect and preprocess multi-dimensional data and build lightweight intelligent analysis logic for core operation and maintenance scenarios.

[0013] The cross-site resource coordination unit is used to manage maintenance personnel, spare parts and work orders in a unified manner, and intelligently generate the optimal work order dispatch plan;

[0014] The full lifecycle management unit is used to establish an electronic health record for each charging pile, track equipment status, predict lifespan, and generate preventive maintenance plans.

[0015] The report generation unit is used to integrate data from across the entire domain to generate analytical reports;

[0016] The communication management unit is used to manage cloud-edge communication links, set data transmission priorities, and complete offline data reception, verification, and storage.

[0017] Preferably, the big data analysis unit has the following specific workflow:

[0018] Collect and preprocess multi-dimensional data uploaded by the terminal device's perception and execution module and the distributed edge collaboration module;

[0019] Four major operation and maintenance scenarios were defined: fault prediction, load forecasting, equipment health assessment, and operation and maintenance strategy optimization. For each scenario, based on the charging pile operation mechanism and operation and maintenance management requirements, strong correlation core features were extracted from the preprocessed data.

[0020] Fault prediction is built using a fusion of LSTM (Long Short-Term Memory) network and random forest algorithm; load prediction is built using time series prediction algorithm and gradient boosting tree; health assessment is built using analytic hierarchy process and fuzzy comprehensive evaluation method; and operation and maintenance strategy optimization is built using reinforcement learning algorithm.

[0021] The algorithm is trained and fitted using full-domain data and the parameters are adjusted through cross-validation. The algorithm is then adapted to be lightweight, forming intelligent analysis logic that can run locally at the edge. The lightweight logic and parameters are periodically distributed to the distributed edge collaboration module.

[0022] Preferably, the full lifecycle management unit specifically includes: assigning a unique identifier to each device and establishing a health record; synchronizing runtime, number of charging cycles, electrical parameters, temperature, and fault information in real time and updating the record; automatically generating a preventive maintenance plan based on health assessment and lifespan prediction; and automatically providing scrapping suggestions based on service life, health status, maintenance costs, and fault frequency to assist in asset renewal decisions.

[0023] Preferably, the cross-site resource coordination unit specifically includes: real-time location of personnel status, skills and work order progress, and intelligent matching of the best maintenance personnel; establishment of a unified inventory ledger to support inventory monitoring and cross-regional transfer early warning; online management of work order creation, dispatch, tracking, feedback and archiving, while classifying fault levels and priorities.

[0024] Preferably, the global computing power scheduling unit specifically includes:

[0025] Local tasks with extremely high real-time requirements and small data volume are designated as edge-specific tasks and are all handled by the local computing power of the corresponding distributed edge collaboration module for nearby processing.

[0026] Non-real-time tasks, large data volumes, and tasks requiring high computing power are designated as exclusive tasks of the cloud-based overall management and control module, and all of them are centrally processed by the cloud-based overall management and control module.

[0027] By offloading computing power in a tiered manner, the computing power load of the cloud-based overall management and control module and the distributed edge collaboration module is balanced, thereby reducing network bandwidth consumption.

[0028] Preferably, the distributed edge collaboration module includes multiple distributed edge collaboration sub-modules, and a single distributed edge collaboration sub-module specifically includes:

[0029] The local real-time control unit is used to realize local charging control, emergency fault handling and status monitoring;

[0030] The edge inference unit is used to load lightweight intelligent analysis logic to complete fault prediction, load forecasting, health assessment and anomaly identification.

[0031] The data caching unit is used to preprocess the raw data and cache key operation, fault, and model parameters;

[0032] The offline autonomous unit is used to monitor the cloud-edge communication status. It automatically enters offline autonomy when the network is lost and automatically synchronizes data after communication is restored.

[0033] Edge communication unit, used to enable communication between the edge and the terminal, the cloud-based overall management and control module, and adjacent edge nodes;

[0034] Cross-edge collaboration unit is used to establish linkage with adjacent edge nodes, execute load sharing, redundancy switching, and work order cross-regional transfer instructions.

[0035] Preferably, the edge reasoning unit specifically includes:

[0036] Load lightweight intelligent analysis logic to perform fault, load, health and abnormal behavior analysis locally, and quickly trigger alarms and control commands;

[0037] The data on operational deviations, fault samples, analysis errors, and handling results are transmitted back to the cloud-based integrated management module.

[0038] The cloud-based overall management and control module optimizes algorithm parameters, analysis logic, and feature library based on feedback data, and updates lightweight rules;

[0039] The iteratively updated lightweight intelligent analysis logic is periodically distributed to each distributed edge collaboration submodule to achieve continuous algorithm upgrades.

[0040] Preferably, the cross-edge node cluster linkage submodule specifically includes:

[0041] When the power grid load in a region exceeds the limit, the power of adjacent low-load edge nodes will be adjusted or the charging demand will be diverted.

[0042] When the distributed edge collaboration submodule fails or there is a large-scale failure in the managed stations, the adjacent nodes will be linked to initiate redundant takeover to ensure that charging and maintenance are not interrupted.

[0043] When local maintenance resources are insufficient, work orders will be transferred across regions to adjacent nodes to coordinate personnel and spare parts support.

[0044] Preferably, the fault redundancy switching unit specifically includes:

[0045] Real-time monitoring of the status of each distributed edge collaboration sub-module and terminal device, and collection of hardware, communication and large-scale site fault information;

[0046] Verify the fault level and initiate redundancy switching when a complete breakdown or uncontrollable, or a series of faults cannot be handled.

[0047] Based on geographical location, control range and number of terminals, intelligently match adjacent normal and computing-powered edge distributed edge collaboration sub-modules as takeover parties and issue takeover instructions;

[0048] Take over the distributed edge collaboration submodule, remove control restrictions, start cross-regional redundancy mode, load parameters and policies, and switch communication links;

[0049] During the takeover period, the team was responsible for the full-process operation and maintenance of the faulty area and transmitted data back in real time.

[0050] After troubleshooting, the cloud-based overall management module issues a switchback command, takes over the distributed edge collaboration submodule, restores permissions, completes data alignment, and restores the original cloud-edge collaboration architecture.

[0051] Real-time monitoring of the online status and operating conditions of each distributed edge collaboration submodule and its subordinate site terminal equipment, and synchronous collection of hardware, communication and large-area equipment failure information in the site.

[0052] Compared with the prior art, the beneficial effects of the present invention are:

[0053] 1. This invention constructs a cloud-edge collaborative system that combines global coordination and regional autonomy, taking into account the global decision-making advantages of the cloud-based overall management and control module with the real-time response capabilities of the edge. It completes high-real-time fault handling and local control in milliseconds, while non-real-time big data analysis and strategy optimization are handled by the cloud-based overall management and control module, thereby achieving optimal computing power configuration and significantly reducing network bandwidth usage and operating latency.

[0054] 2. This invention has offline autonomy and dual redundancy protection capabilities. After the edge node is disconnected from the network, it can independently complete all operation and maintenance management business. The dual redundancy design of hardware and cloud edge avoids the failure of a single module from causing the site to shut down, ensuring the continuous and stable operation of a large-scale charging pile group and significantly improving the reliability of operation.

[0055] 3. This invention has the ability to interact with the power grid across edge clusters, realize load sharing, fault redundancy switching, orderly charging and power grid peak shaving and valley filling, balance the power grid load, adapt to virtual power plants and integrated source-grid-load-storage application scenarios, and broaden the application value of charging pile clusters. Attached Figure Description

[0056] Figure 1 This invention relates to a cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters.

[0057] Figure 2 This invention relates to a cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters.

[0058] Figure 3 This is a schematic diagram of the cloud-based overall management and control module of the present invention;

[0059] Figure 4 This is a schematic diagram of the cross-edge node cluster linkage submodule of the present invention. Detailed Implementation

[0060] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0061] To address the pain points of existing technologies, such as lack of cloud-edge collaboration, network outage paralysis, algorithm rigidity, lack of resource coordination, and susceptibility to power grid imbalance, which make them unsuitable for intelligent operation and maintenance, please refer to [link / reference needed]. Figures 1-4 This embodiment provides the following technical solution:

[0062] A cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters includes:

[0063] The cloud-based overall management and control module is used to realize global computing power scheduling, full-scale big data analysis, intelligent algorithm iteration and optimization, global operation and maintenance strategy generation, cross-site resource coordination, and full life cycle asset management. Unlike the traditional system's simple data collection end, it has the core capabilities of global collaborative management and intelligent decision-making, and coordinates the operation and maintenance decisions and resource scheduling of the entire cluster.

[0064] The distributed edge collaboration module is used to realize local real-time control, millisecond-level fault response, local computing power offloading, multi-source data preprocessing and caching functions;

[0065] The terminal device sensing and execution module is used to collect charging pile operating parameters, environmental data, charging status and fault information in real time, receive control commands issued by the distributed edge collaboration module and execute them accurately, and complete basic execution actions such as charging start and stop, power adjustment, fault reporting and status feedback.

[0066] The cross-edge node cluster linkage sub-module is used to realize the pioneering cross-distributed edge collaborative module collaborative mechanism and to complete the mutual support of charging pile load, fault redundancy switching and cross-regional transfer of operation and maintenance work orders between the distributed edge collaborative modules of adjacent stations.

[0067] The cloud-based overall management and control module specifically includes:

[0068] The global computing power scheduling unit is used to monitor the computing power load of the cloud-based overall management and control module and the distributed edge collaboration module in real time, establish computing power hierarchical scheduling rules, and dynamically allocate computing power resources between the cloud-based overall management and control module and the edge side according to the task's real-time level, data volume, and computing power requirements to achieve computing power load balancing.

[0069] The big data analysis unit is used to collect all historical operation data, fault samples, real-time operating data, power grid load linkage data, and charging demand fluctuation data uploaded by the perception and execution module of the entire cluster terminal equipment and each distributed edge collaboration sub-module. After data cleaning, feature extraction, and normalization preprocessing to remove redundant noise information, it relies on deep learning algorithms, machine learning algorithms and big data analysis engine to build intelligent analysis logic step by step for core operation and maintenance scenarios.

[0070] The cross-site resource coordination unit is used to uniformly manage the entire cluster's maintenance personnel, spare parts, and work orders. Based on the fault location, fault level, real-time location of maintenance personnel, and spare parts inventory distribution, it intelligently generates the optimal work order dispatch plan to achieve global optimization of maintenance resources.

[0071] The full lifecycle management unit is used to establish an independent electronic health record for the terminal device sensing and execution module corresponding to each terminal charging pile device, track the device's status throughout the entire process, predict the device's remaining service life, and generate a preventive maintenance plan. The electronic health record includes: recording device installation information, operating parameters, fault records, maintenance records, and spare parts replacement information. It tracks the device's status throughout the entire process from installation, operation, maintenance to scrapping, predicts the device's remaining service life, and generates a preventive maintenance plan.

[0072] The report generation unit is used to integrate relevant data from the entire cluster, including operational data, maintenance data, load data, and fault data, and generate corresponding analysis reports. These reports include multi-dimensional visualization reports, maintenance analysis reports, and trend prediction reports, providing data support for operation management and decision-making.

[0073] The communication management unit is used to manage the communication link between the cloud-based overall control module and the distributed edge collaboration module, ensure communication stability, set data transmission priorities, prioritize the transmission of critical data such as fault alarms and control commands, and simultaneously receive, verify, and synchronize offline data from the distributed edge collaboration module into the database.

[0074] The big data analytics unit's specific workflow is as follows:

[0075] The system collects and preprocesses multi-dimensional data uploaded by the terminal device's perception and execution module and the distributed edge collaboration module. The multi-dimensional data includes: full historical operation data, fault samples, real-time operating condition data, power grid interaction data, power grid load linkage data, and charging demand fluctuation data. The preprocessing includes: data cleaning, feature extraction, and normalization preprocessing.

[0076] Four major operation and maintenance scenarios were defined: fault prediction, load forecasting, equipment health assessment, and operation and maintenance strategy optimization. For each scenario, based on the charging pile operation mechanism and operation and maintenance management requirements, strong correlation core features were extracted from the preprocessed data. Specifically, the fault prediction scenario extracted core features such as voltage, current, temperature, insulation impedance, and abnormal alarms; the load forecasting scenario extracted core features such as regional power grid load curves, time-period charging demand, historical charging data, and travel peak correlations; the health assessment scenario extracted core features such as running time, charging frequency, component wear, and operation deviations; and the operation and maintenance strategy optimization scenario extracted core features such as fault handling efficiency, resource consumption, operation and maintenance costs, and customer satisfaction. Feature screening and normalization were completed to remove noise and invalid data.

[0077] Fault prediction is built using a fusion of LSTM (Long Short-Term Memory) network and random forest algorithm; load prediction is built using time series prediction algorithm and gradient boosting tree; health assessment is built using analytic hierarchy process and fuzzy comprehensive evaluation method; and operation and maintenance strategy optimization is built using reinforcement learning algorithm.

[0078] Based on preprocessed massive data across the entire domain, various algorithms are trained and fitted. Through cross-validation, the algorithm parameters are repeatedly debugged and the analysis deviations are corrected to ensure that the accuracy of fault prediction, the precision of load forecasting, and the reliability of health assessment meet the standards and meet the requirements of large-scale charging pile group operation and maintenance management.

[0079] The optimized algorithms undergo lightweight adaptation processing, which includes: trimming redundant computational logic, simplifying the operation process, compressing parameter size, and eliminating invalid branches. This reduces the algorithm's computing power and storage requirements, adapts it to the low computing power, low storage, and low power consumption hardware operating environment at the edge, and forms a lightweight intelligent analysis logic that can run locally at the edge.

[0080] After completing the full-process debugging and accuracy verification, the lightweight intelligent analysis logic and supporting operating parameters are regularly distributed to the distributed edge collaboration module to ensure the accuracy and efficiency of local analysis on the edge side, without relying on the real-time computing power support of the cloud-based overall management module.

[0081] The full lifecycle management unit enables closed-loop control of the entire process for large-scale charging pile assets, specifically including:

[0082] Each terminal charging pile device is assigned a unique identifier, and basic information such as device model, manufacturer, installation time, installation location, and technical parameters is entered to establish a unique health record, enabling full life-cycle traceability of a single device;

[0083] Real-time synchronization of device runtime, charging cycles, voltage and current parameters, temperature data, number of faults and fault types, and continuous updating of device health records;

[0084] Based on equipment health assessment results and remaining life prediction data, a personalized preventive maintenance plan is automatically generated, specifying maintenance time, maintenance content, and maintenance spare parts, and reminding maintenance personnel to perform maintenance work in advance, rather than traditional passive fault repair;

[0085] By combining equipment operating years, health status, maintenance costs, and failure frequency, the system automatically assesses the scrap value of equipment, generates scrapping recommendations, assists in asset renewal decisions, achieves refined asset management, maximizes equipment lifespan, and reduces asset idleness and waste.

[0086] The cross-site resource coordination unit enables comprehensive and refined management and control of operation and maintenance resources, specifically including:

[0087] Real-time location tracking of all maintenance personnel across the entire cluster, including their on-duty status, skill level, and work order processing progress. Establish dynamic personnel profiles and intelligently match the optimal maintenance personnel based on fault location and severity to avoid personnel being idle or overworked.

[0088] Establish a unified ledger for spare parts inventory across the entire cluster, and update the quantity, storage location, and loss status of spare parts inventory in each site and region in real time. Support cross-regional spare parts allocation and early warning. When the spare parts inventory in a certain region is insufficient, automatically link with surrounding areas for allocation to ensure the supply of spare parts for fault repair.

[0089] It enables online management of the entire process of work order creation, intelligent dispatching, progress tracking, result feedback, and acceptance archiving. It supports fault level classification, work order priority division, priority handling of emergency fault work orders, and full recording of work order processing trajectory, which facilitates operation and maintenance assessment and review optimization.

[0090] The global computing power scheduling unit establishes a tiered computing power offloading mechanism to accurately distinguish between real-time and non-real-time tasks, achieving optimal computing power allocation. Specifically, this includes:

[0091] Local tasks such as millisecond-level emergency stop control, short circuit warning, local fault alarm, real-time charging control, and local security protection are designated as edge-specific tasks. All of these tasks are handled by the local computing power of the corresponding distributed edge collaboration module, ensuring a response latency of less than 10ms and preventing network latency from affecting operational security.

[0092] The tasks of full cluster load trend analysis, equipment life prediction, long-term operation and maintenance planning, big data report generation, global strategy optimization, and cross-site collaborative scheduling are designated as exclusive tasks of the cloud-based overall management and control module. All of these tasks are centrally processed by the cloud-based overall management and control module to avoid wasting the computing power of the distributed edge collaborative module.

[0093] By offloading computing power in a tiered manner, the computing load of the cloud-based overall management and control module and the distributed edge collaboration module is balanced, reducing the network bandwidth consumption caused by the long-distance transmission of massive amounts of data, improving the overall system operating efficiency and response speed, and ensuring the reliability of high real-time tasks.

[0094] The distributed edge collaboration module comprises multiple distributed edge collaboration sub-modules, and a single distributed edge collaboration sub-module specifically includes:

[0095] The local real-time control unit is used to receive real-time operating data uploaded by the sensing and execution modules of terminal devices within the corresponding control range, and to perform local charging control, emergency fault handling and equipment status monitoring functions to achieve millisecond-level fault response and ensure the safe operation of local stations.

[0096] The edge inference unit is used to load lightweight intelligent analysis logic issued by the cloud-based overall management and control module to complete local fault prediction, load forecasting, real-time equipment health assessment, and abnormal behavior identification and inference. The inference results are directly used for local operation and maintenance decisions and alarm triggering.

[0097] The data caching unit is used to filter, clean, convert formats, and extract features from the raw data uploaded by the terminal device's perception and execution module. It removes redundant data and uploads only the core and valid data to the cloud-based overall management and control module. At the same time, it caches local key operating data, fault data, and model parameters to ensure data support for local task execution.

[0098] The offline autonomous unit is used to detect the communication status between the cloud-based overall management and control module and the distributed edge collaboration sub-module. When a network interruption or communication failure is detected, it automatically switches to offline autonomous mode, enables local data caching and lightweight intelligent analysis logic, and independently completes all operation and maintenance management, charging control and fault response work of the local site without affecting the normal operation of the terminal charging pile equipment. When communication is restored, the data synchronization mechanism is automatically triggered to upload all operating data, fault records and operation and maintenance records during the offline period to the cloud-based overall management and control module, and complete the alignment of policy parameters with the cloud-based overall management and control module.

[0099] The edge communication unit is used to realize local communication between the distributed edge collaboration submodule and the terminal device perception and execution module, remote communication between the distributed edge collaboration submodule and the cloud-based overall management and control module, and collaborative communication between the distributed edge collaboration submodule and adjacent distributed edge collaboration submodules.

[0100] The cross-edge collaboration unit is used to establish collaborative communication links with the distributed edge collaboration sub-modules in adjacent areas, execute load sharing instructions, fault redundancy switching instructions, and cross-regional work order transfer instructions to achieve cluster-level collaborative operation and maintenance.

[0101] Edge reasoning units, specifically including:

[0102] The distributed edge collaboration submodule loads lightweight intelligent analysis logic, without relying on the cloud-based overall management module's computing power and network connection. It completes the entire process of fault prediction, load forecasting, equipment health assessment, and abnormal behavior identification locally. The analysis results directly trigger local alarms, equipment control adjustments, or maintenance commands, ensuring real-time local response speed.

[0103] During local analysis, the distributed edge collaboration submodule synchronously collects various types of effective feedback data, including actual operational deviation data, newly added fault samples, analysis error data, and on-site handling results data. It then transmits the above data back to the cloud-based overall management and control module, providing real-scenario data support for the algorithm iteration of the cloud-based overall management and control module.

[0104] The cloud-based overall management and control module, based on the new data transmitted back from the edge side, fine-tunes the parameters, optimizes the logic, and updates the features of the original lightweight intelligent analysis logic. It also optimizes the lightweight adaptation rules on the edge side to improve the accuracy of algorithm analysis and the adaptability of scenarios.

[0105] The newly optimized and iterated version of the lightweight intelligent analysis logic is regularly distributed to each distributed edge collaboration sub-module to replace the old version. This forms a complete closed loop of data processing in the cloud-based overall management module, local edge analysis, data feedback, and algorithm iteration in the cloud-based overall management module. This continuously improves the accuracy of system operation and maintenance management, achieves the technical effect of the system becoming smarter with use, and completely solves the technical defects of traditional edge system analysis logic being static, lacking prediction accuracy, and unable to be dynamically iterated.

[0106] The cross-edge node cluster linkage submodule specifically includes:

[0107] The cloud-based overall management and control module monitors the power grid load and charging pile power load of each area in real time. When the power grid load of a certain area approaches the threshold and the charging pile is running under power restriction, the cloud-based overall management and control module issues a coordination instruction to link the distributed edge coordination module of the adjacent low-load area, coordinate the charging piles of the adjacent sites to appropriately divert charging demand or adjust the charging power, so as to achieve cross-area load balance, avoid overload of the area power grid, and ensure the continuity of charging service.

[0108] When a large-scale failure occurs in a distributed edge collaboration module or its management site, and it cannot operate normally, the cloud-based overall management module issues a switching command. The adjacent distributed edge collaboration module automatically starts the redundant management mode and temporarily takes over the basic operation and maintenance and charging control functions of some or all terminal equipment sensing and execution modules in the faulty area, so as to achieve seamless fault switching and prevent large-scale outages.

[0109] When a large number of failures occur in a certain area, or when there is a shortage of local maintenance personnel or spare parts, the cloud-based overall management and control module coordinates maintenance resources in surrounding areas and transfers some maintenance work orders across regions to the distributed edge collaboration module in adjacent areas. This coordinates the support of surrounding maintenance personnel and spare parts, enabling rapid completion of fault handling and improving overall fault response efficiency.

[0110] The fault redundancy switching unit specifically includes:

[0111] Real-time monitoring of the online status and operating conditions of each distributed edge collaboration submodule and the operating status of terminal equipment in the field under its jurisdiction; synchronous collection of hardware failure, communication failure and large-area equipment failure information uploaded by the distributed edge collaboration submodule.

[0112] Verify the fault level to confirm that the target distributed edge collaborative submodule has been completely paralyzed and lost its autonomous control capability, or that the sites under its jurisdiction have experienced a series of faults that cannot be handled autonomously, and meet the fault redundancy switching start conditions.

[0113] Based on the geographical location, control range, and number of terminal devices of the faulty distributed edge collaboration sub-module, an adjacent distributed edge collaboration sub-module that is in normal operation and has sufficient computing power and load margin is intelligently matched as the redundant takeover subject.

[0114] Clearly define the scope of takeover, takeover authority, operation and maintenance management content, operating parameter configuration, and switchover time, and simultaneously issue the instructions to the faulty distributed edge collaboration submodule and the adjacent redundant takeover distributed edge collaboration submodule to ensure accurate delivery of instructions.

[0115] After receiving the switching instruction from the cloud-based overall management and control module, the adjacent redundant takeover distributed edge collaboration submodule automatically closes the original local management and control restrictions, starts the cross-regional redundant management and control mode, and quickly loads the core operation and maintenance parameters, equipment configuration information and management and control policies of the faulty area synchronized by the cloud-based overall management and control module.

[0116] Simultaneously complete the communication link switch, establish a stable point-to-point connection with the terminal equipment perception and execution module in the fault area, complete the seamless handover of control authority, and ensure that the basic charging and safety protection functions of the terminal charging pile are not interrupted throughout the process.

[0117] During the takeover operation monitoring and fault recovery switchover redundancy takeover, the adjacent distributed edge collaborative sub-modules undertake the entire process of terminal equipment monitoring, charging control, fault alarm and basic operation and maintenance in the faulty area. At the same time, the takeover operation data is transmitted back to the cloud-based overall management and control module in real time. The cloud-based overall management and control module monitors the takeover operation status in real time to ensure stable management and control.

[0118] Once the faulty distributed edge collaborative submodule or its subordinate site has been troubleshooted and restored to normal operation, the cloud-based overall management and control module issues a switchback command. The adjacent takeover distributed edge collaborative submodule closes its redundant management and control mode, returns the operation and maintenance management authority to the faulty distributed edge collaborative submodule, and simultaneously completes data and parameter alignment, restoring the original cloud-edge collaborative management and control pattern, thus achieving a closed-loop process for fault redundancy switching.

[0119] Working principle: When using the cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile groups of this invention, according to... Figures 1-4 This includes the following steps:

[0120] S1: The global computing power scheduling unit distinguishes between real-time and non-real-time tasks. Real-time tasks are handled locally at the edge, while non-real-time big data tasks are handled centrally by the cloud-based overall management and control module to achieve computing power load balancing.

[0121] S2: The cloud-based overall management and control module's big data analysis unit relies on full-domain and full-volume data to filter features according to scenarios, build algorithms, complete accuracy optimization and lightweight adaptation, form lightweight intelligent analysis logic that can be used on the edge side, and distribute it to the distributed edge collaboration sub-module;

[0122] S3: The distributed edge collaboration submodule loads lightweight intelligent analysis logic to perform local real-time analysis, synchronously collects and analyzes deviation data and new fault samples and sends them back to the cloud-based overall management and control module. The cloud-based overall management and control module continuously iterates and optimizes the algorithm based on the returned data, updates the version of the lightweight analysis logic, forms a closed loop iteration, and continuously improves the accuracy of operation and maintenance analysis and prediction.

[0123] S4: The distributed edge collaboration submodule monitors the status of terminal devices in real time, performs millisecond-level fault response, switches to offline autonomous mode when the network is interrupted, and automatically completes data and parameter synchronization after reconnection.

[0124] S5: Relying on the cross-edge node cluster linkage sub-module, it performs cross-regional load sharing, fault redundancy switching, and cross-regional transfer of operation and maintenance work orders to achieve cluster collaborative operation and maintenance;

[0125] S6: Through the cross-site resource coordination unit, it achieves comprehensive and refined management and intelligent allocation of maintenance personnel, spare parts, and work orders, thereby improving the efficiency of maintenance and handling.

[0126] S7: Based on the full lifecycle management unit, establish equipment health records, generate preventive maintenance plans, and realize full-process traceability and proactive operation and maintenance of equipment.

[0127] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0128] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

Claims

1. A cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters, characterized in that: include: The cloud-based overall management and control module is used to realize global computing power scheduling, full-scale big data analysis, intelligent algorithm iteration, global operation and maintenance strategy generation, cross-site resource coordination, and full life cycle asset management; The distributed edge collaboration module is used to realize local real-time control, millisecond-level fault response, local computing power offloading, multi-source data preprocessing and caching functions; The terminal device sensing and execution module is used to collect charging pile operating parameters, environmental data, charging status and fault information in real time, receive control commands issued by the distributed edge collaboration module and execute them accurately, and complete basic execution actions such as charging start and stop, power adjustment, fault reporting and status feedback. The cross-edge node cluster linkage submodule is used to collect charging pile operating parameters, environmental data, charging status and fault information in real time, and to perform charging start and stop, power adjustment, fault reporting and status feedback.

2. The cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters according to claim 1, characterized in that, The cloud-based overall management and control module specifically includes: The global computing power scheduling unit is used to monitor the cloud-based overall management and control module and the edge computing power load in real time, establish hierarchical scheduling rules, and dynamically allocate computing power resources; The big data analytics unit is used to collect and preprocess multi-dimensional data and build lightweight intelligent analysis logic for core operation and maintenance scenarios. The cross-site resource coordination unit is used to manage maintenance personnel, spare parts and work orders in a unified manner, and intelligently generate the optimal work order dispatch plan; The full lifecycle management unit is used to establish an electronic health record for each charging pile, track equipment status, predict lifespan, and generate preventive maintenance plans. The report generation unit is used to integrate data from across the entire domain to generate analytical reports; The communication management unit is used to manage cloud-edge communication links, set data transmission priorities, and complete offline data reception, verification, and storage.

3. The cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters according to claim 2, characterized in that, The specific workflow of the big data analysis unit is as follows: Collect and preprocess multi-dimensional data uploaded by the terminal device's perception and execution module and the distributed edge collaboration module; Four major operation and maintenance scenarios were defined: fault prediction, load forecasting, equipment health assessment, and operation and maintenance strategy optimization. For each scenario, strong correlation core features were extracted from the preprocessed data in combination with the charging pile operation mechanism and operation and maintenance management requirements. Fault prediction is built using a fusion of LSTM (Long Short-Term Memory) network and random forest algorithm; load prediction is built using time series prediction algorithm and gradient boosting tree; health assessment is built using analytic hierarchy process and fuzzy comprehensive evaluation method; and operation and maintenance strategy optimization is built using reinforcement learning algorithm. The algorithm is trained and fitted using full-domain data and the parameters are adjusted through cross-validation. The algorithm is then adapted to be lightweight, forming intelligent analysis logic that can run locally at the edge. The lightweight logic and parameters are periodically distributed to the distributed edge collaboration module.

4. The cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters according to claim 2, characterized in that, The full lifecycle management unit specifically includes: assigning a unique identifier to each device and establishing a health record; synchronizing runtime, number of charging cycles, electrical parameters, temperature, and fault information in real time and updating the record; automatically generating preventive maintenance plans based on health assessments and lifespan predictions; and automatically providing scrapping recommendations based on service life, health status, maintenance costs, and fault frequency to assist in asset replacement decisions.

5. The cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters according to claim 2, characterized in that, The cross-site resource coordination unit specifically includes: real-time location tracking of personnel status, skills, and work order progress; intelligent matching of optimal maintenance personnel; establishment of a unified inventory ledger to support inventory monitoring and cross-regional transfer early warning; online management of work order creation, dispatch, tracking, feedback, and archiving; and classification of fault levels and priorities.

6. The cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters according to claim 2, characterized in that, The global computing power scheduling unit specifically includes: Local tasks with extremely high real-time requirements and small data volume are designated as edge-specific tasks and are all handled by the local computing power of the corresponding distributed edge collaboration module for nearby processing. Non-real-time tasks, large data volumes, and tasks requiring high computing power are designated as exclusive tasks of the cloud-based overall management and control module, and all of them are centrally processed by the cloud-based overall management and control module. By offloading computing power in a tiered manner, the computing power load of the cloud-based overall management and control module and the distributed edge collaboration module is balanced, thereby reducing network bandwidth consumption.

7. The cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters according to claim 1, characterized in that, The distributed edge collaboration module includes multiple distributed edge collaboration sub-modules, and a single distributed edge collaboration sub-module specifically includes: The local real-time control unit is used to realize local charging control, emergency fault handling and status monitoring; The edge inference unit is used to load lightweight intelligent analysis logic to complete fault prediction, load forecasting, health assessment and anomaly identification. The data caching unit is used to preprocess the raw data and cache key operation, fault, and model parameters; The offline autonomous unit is used to monitor the cloud-edge communication status. It automatically enters offline autonomy when the network is lost and automatically synchronizes data after communication is restored. Edge communication unit, used to enable communication between the edge and the terminal, the cloud-based overall management and control module, and adjacent edge nodes; Cross-edge collaboration unit is used to establish linkage with adjacent edge nodes, execute load sharing, redundancy switching, and work order cross-regional transfer instructions.

8. The cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters according to claim 7, characterized in that, The edge reasoning unit specifically includes: Load lightweight intelligent analysis logic to perform fault, load, health and abnormal behavior analysis locally, and quickly trigger alarms and control commands; The data on operational deviations, fault samples, analysis errors, and handling results are transmitted back to the cloud-based integrated management module. The cloud-based overall management and control module optimizes algorithm parameters, analysis logic, and feature library based on feedback data, and updates lightweight rules; The iteratively updated lightweight intelligent analysis logic is periodically distributed to each distributed edge collaboration submodule to achieve continuous algorithm upgrades.

9. The cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters according to claim 1, characterized in that, The cross-edge node cluster linkage submodule specifically includes: When the power grid load in a region exceeds the limit, the power of adjacent low-load edge nodes will be adjusted or the charging demand will be diverted. When the distributed edge collaboration submodule fails or there is a large-scale failure in the managed stations, the adjacent nodes will be linked to initiate redundant takeover to ensure that charging and maintenance are not interrupted. When local maintenance resources are insufficient, work orders will be transferred across regions to adjacent nodes to coordinate personnel and spare parts support.

10. The cloud-edge collaborative intelligent operation and maintenance management system for large-scale charging pile clusters according to claim 9, characterized in that, The fault redundancy switching unit specifically includes: Real-time monitoring of the status of each distributed edge collaboration sub-module and terminal device, and collection of hardware, communication and large-scale site fault information; Verify the fault level and initiate redundancy switching when a complete breakdown or uncontrollable, or a series of faults cannot be handled. Based on geographical location, control range and number of terminals, intelligently match adjacent normal and computing-powered distributed edge collaboration sub-modules as takeover parties and issue takeover instructions; Take over the distributed edge collaboration submodule, remove control restrictions, start cross-regional redundancy mode, load parameters and policies, and switch communication links; During the takeover period, the team was responsible for the full-process operation and maintenance of the faulty area and transmitted data back in real time. After troubleshooting, the cloud-based overall management module issues a switchback command, takes over the distributed edge collaboration submodule, restores permissions, completes data alignment, and restores the original cloud-edge collaboration architecture. Real-time monitoring of the online status and operating conditions of each distributed edge collaboration submodule and its subordinate site terminal equipment, and synchronous collection of hardware, communication and large-area equipment failure information in the site.

Images