A method for operation and maintenance of a charging station based on a multi-dimensional data system

Abstract

The present invention discloses a method for the operation and maintenance of a charging station based on a multi-dimensional data system, which relates to the field of intelligent optimization. It includes collecting data on the operation of charging piles, the environment, users, and the power grid. After preprocessing it into JSON, it is parsed and classified in real time through Kafka-Flume, stored in a distributed database, and a composite index is established, and retrieved through Presto. After cleaning, a spatio-temporal map is constructed by combining GraphQL and graph query language, and multi-dimensional feature vectors are extracted. A digital twin model is generated, and multi-source data is fused through virtual-real mapping, and combined with residual correction and reinforcement learning for optimization. A hypergraph neural network is used to fuse multi-modal data, and combined with an entropy regularization multi-objective strategy and meta-learning to optimize the topology. Power is dynamically allocated, LSTM-GAN is used to predict and locate faults, and PID is used to regulate the output. The blockchain contract automatically triggers equipment freezing and power limiting, and the data on the chain is fed back to optimize scheduling and load balancing. The present invention significantly improves the accuracy and response efficiency of the operation and maintenance of the charging station through a multi-dimensional data fusion and dynamic optimization mechanism.

Description

Technical Field

[0001] The present invention relates to the field of intelligent optimization technology, and particularly to a method for operation and maintenance of a charging station based on a multi-dimensional data system. Background Art

[0002] Currently, the operation and maintenance technology of charging stations mainly relies on Internet of Things sensors to collect the operation data of charging piles, and performs offline analysis through batch processing frameworks such as Hadoop, for example, statistical modeling and rule-based early warning based on historical data. Although the existing solutions can achieve basic status monitoring, the integration efficiency of multi-source heterogeneous data is low. The traditional ETL process relies on the association of static data tables and lacks the ability of real-time parsing and dynamic classification. Especially in the scenario of cross-dimensional data fusion, the spatio-temporal correlation analysis is limited by a single indexing mechanism, resulting in the coupling relationship between environmental factors and user behavior on the power grid load being difficult to mine, forming a data island effect.

[0003] The main defect of the existing technology lies in the lack of real-time integration and dynamic association capabilities of multi-source heterogeneous data. The traditional architecture cannot efficiently process high-concurrency streaming data and lacks composite index support, resulting in high data retrieval latency and coarse-grained cross-dimensional analysis. For example, when the power grid load suddenly changes, it is impossible to real-time associate the charging behavior characteristics of users with environmental parameters, making the power distribution strategy lag, and it is difficult to balance the charging efficiency and the stability of the power grid. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for operation and maintenance of a charging station based on a multi-dimensional data system to solve the problems of low dynamic integration efficiency of multi-source heterogeneous data of the charging station, real-time regulation deviation caused by insufficient adaptive optimization of the digital twin model, and weak generalization ability of fault prediction.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for operation and maintenance of a charging station based on a multi-dimensional data system, which includes collecting multi-dimensional data of the charging station, packaging it into a JSON format data packet and performing preprocessing, where the multi-dimensional data includes charging pile operation data, environmental data, user data, and power grid load data;

[0008] Using a Kafka-Flume combined architecture to parse and classify the JSON format data packet, storing it in a database and establishing a composite index, and using Presto query to perform data union retrieval and output a retrieval result set;

[0009] Perform multi-dimensional standardization processing on the retrieval result set, dynamically associate multi-dimensional data using a graph query language, construct a spatio-temporal association graph, extract multi-dimensional data relationship features, and generate multi-dimensional feature vectors;

[0010] Based on the multi-dimensional feature vectors, combine computer vision to generate a 3D model of the charging station, use a virtual-real mapping equation to dynamically fuse multi-dimensional data into the 3D model, and construct a digital twin model;

[0011] Calculate the residual error between the digital twin model and the real data of the charging station in real time, and correct the digital twin model;

[0012] Combine the reinforcement learning algorithm to optimize the interaction efficiency, and dynamically adjust the digital twin model through adaptive gain,

[0013] Use a hypergraph neural network to fuse multi-dimensional feature vectors, combine an entropy-regularized multi-objective MDP update strategy, and use a meta-learning framework to quickly adapt to different tasks, and dynamically optimize the hypergraph topology;

[0014] Perform power allocation through a dynamic optimization algorithm, use LSTM-GAN to fuse and predict faults and locate them, and feedback cross-scenario optimization of resource scheduling and load balancing.

[0015] As a preferred solution of the charging station operation and maintenance method based on the multi-dimensional data system of the present invention, wherein: the packaging into a JSON format data packet and preprocessing include,

[0016] Use the Kalman filter algorithm to remove noise, detect anomalies through the Z-Score method, and use the linear interpolation algorithm for completion.

[0017] As a preferred solution of the charging station operation and maintenance method based on the multi-dimensional data system of the present invention, wherein: the Kafka-Flume combined architecture for parsing and classification, storing in the database and establishing a composite index means,

[0018] Real-time parse the JSON format data packet through Kafka, and dynamically classify it into charging pile operation data, environmental data, user data, grid load data and the original multi-dimensional data of the charging station based on JSON fields;

[0019] Use Flume to store the charging pile operation data, user data and grid load data into a distributed relational database, and establish a time series index, user identification index and geographical location index. The charging station is partitioned and stored in the distributed file system according to the timestamp and device ID.

[0020] As a preferred solution of the charging station operation and maintenance method based on the multi-dimensional data system of the present invention, wherein: the use of graph query language to dynamically associate multi-dimensional data, construct a spatio-temporal association map and extract multi-dimensional data relationship features means that,

[0021] Based on the multi-dimensional data of the charging station, the node types are predefined as charging pile nodes, user nodes, environmental nodes and power grid nodes, and the edge relationship types are defined as user-charging pile charging relationship, charging pile-environment location relationship and charging pile-power grid impact relationship;

[0022] The map is updated in real time through the graph query language. When a new user charging event occurs, the environmental data and the power grid load are dynamically associated to generate a spatio-temporal association path, and a combined index of timestamp and geographical hash is created in the spatio-temporal association map;

[0023] The spatio-temporal clustering algorithm is used to extract the hot spots of the user behavior area. Through the calculation of dynamic relationship weights, the coupling strength characteristics of the charging pile and the environment-power grid are extracted to generate a multi-dimensional feature vector.

[0024] As a preferred solution of the charging station operation and maintenance method based on the multi-dimensional data system of the present invention, wherein: the use of the virtual-real mapping equation to dynamically fuse multi-dimensional data into a three-dimensional model to construct a digital twin model means that,

[0025] By comparing the multi-dimensional data of the charging station collected in real time with the predicted state of the three-dimensional model item by item, the mean absolute error residual is calculated;

[0026] Using the error residual feedback adjustment, compare the multi-dimensional data of the charging station collected in real time with the predicted state of the multi-dimensional data of the real-time charging station of the three-dimensional model item by item, calculate the charging pile heat dissipation coefficient, the charging pile power attenuation factor and the user behavior influence weight, and average the output error residual to perform reverse optimization on the virtual-real mapping equation to construct a digital twin model.

[0027] As a preferred solution of the charging station operation and maintenance method based on the multi-dimensional data system of the present invention, wherein: the use of the meta-learning framework to quickly adapt to different tasks means that,

[0028] Based on the historical environmental data, the common features of multi-modal data are extracted through a hypergraph neural network to optimize the initial policy parameters and generate cross-scenario general policy parameters;

[0029] Use gradient backpropagation to calculate the task-specific loss, update the initial policy parameters and deploy them to the digital twin model to adjust the charging pile power and user diversion guidance in real time.

[0030] As a preferred solution of the charging station operation and maintenance method based on the multi-dimensional data system of the present invention, wherein: the power distribution through the dynamic optimization algorithm includes,

[0031] Based on dynamic optimization of the hypergraph topology, the particle swarm optimization algorithm is adopted. The real-time power demand of the charging piles, the power grid load data and the user behavior data are input, and the optimized power distribution instructions for each charging pile are output.

[0032] As a preferred solution of the charging station operation and maintenance method based on the multi-dimensional data system of the present invention, wherein: the use of LSTM-GAN to fuse and predict faults and locate includes,

[0033] Based on LSTM-GAN, learn the time series law of the device state, predict the future fault probability, generate an adversarial network to simulate abnormal data such as radiator overload and power module attenuation, and enhance the training samples;

[0034] By comparing the feature differences between the actual data and the abnormal data generated by GAN, combined with the GPS coordinates of the charging pile to locate the fault, output the fault type, location coordinates and maintenance priority.

[0035] In a second aspect, the present invention provides a computer device, including a memory and a processor, the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the charging station operation and maintenance method based on the multi-dimensional data system described in the first aspect of the present invention is implemented.

[0036] In a third aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, wherein: when the computer program is executed by the processor, any step of the charging station operation and maintenance method based on the multi-dimensional data system described in the first aspect of the present invention is implemented.

[0037] The beneficial effects of the present invention are as follows: Through multi-dimensional data fusion and dynamic optimization mechanism, the accuracy and response efficiency of the charging station operation and maintenance are significantly improved. Through the virtual-real mapping equation and the residual error feedback mechanism, the parameters of the digital twin model are dynamically corrected, combined with the reinforcement learning algorithm to optimize the adaptive gain, so that the prediction error of the digital twin model is reduced. Further, the LSTM-GAN fusion model uses the generative adversarial network to expand abnormal samples, enhances the recognition ability of rare faults (such as power module attenuation), and improves the positioning accuracy. At the same time, the hypergraph neural network based on the meta-learning framework quickly adapts to different task scenarios, optimizes the power distribution strategy, and realizes the automatic execution of device freezing and power limitation through the blockchain smart contract, ensuring stability under the power grid load fluctuation, and achieving the global optimum of charging station resource scheduling, fault response and grid coordination. Description of the Drawings

[0038] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0039] Figure 1 It is a flowchart of multi-dimensional data collection and preprocessing in Embodiment 1.

[0040] Figure 2 It is a flowchart of dynamic data integration and digital twin modeling in Embodiment 1.

[0041] Figure 3 It is a flowchart of intelligent optimization and fault response in Embodiment 1.

[0042] Figure 4 It is a flowchart of resource scheduling and execution feedback in Embodiment 1. Specific Embodiments

[0043] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific embodiments of the present invention with reference to the drawings in the specification.

[0044] In the following description, many specific details are set forth to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0045] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selectively exclusive embodiment from other embodiments.

[0046] Embodiment 1, referring to Figures 1 to 4 , is the first embodiment of the present invention. This embodiment provides a method for operating and maintaining a charging station based on a multi-dimensional data system, including the following steps:

[0047] S1. Collect multi-dimensional data of the charging station, package it into a JSON format data packet, and perform preprocessing, where the multi-dimensional data includes charging pile operation data, environmental data, user data, and grid load data;

[0048] Specifically, during the operation and maintenance of the charging station, the operation data (voltage, current, power, equipment temperature), environmental data (temperature, humidity, air quality), user data, and grid load data of the charging piles in the charging station are collected in real time, and multi-sensor time synchronization is achieved through the NTP protocol.

[0049] In this embodiment, the device operation data, environmental data, user data, and grid load data are collected in real time by sensors, and the sampling frequency is 1 time per second.

[0050] S1.1. The preprocessing of multi-dimensional data refers to using the Internet of Things gateway device to package the collected raw data into JSON format, and adopting the Kalman filtering algorithm to process the raw data, removing the noise introduced by sensor errors or environmental factors, detecting outliers through the Z-Score method, and eliminating the data that does not conform to the conventional change range. If data loss is found, the linear interpolation algorithm is used for supplementation, and the data packet is transmitted to the cloud platform through the 5G network.

[0051] S2. Parse and classify the JSON format data packets using the Kafka-Flume combined architecture, store them in the database and establish a composite index, and use Presto query to perform data joint retrieval and output the retrieval result set.

[0052] It includes the following steps:

[0053] S2.1. Parse the JSON format data packets transmitted to the cloud platform through Kafka (streaming data processing platform). After parsing, the preprocessed multi-dimensional data of the charging station is dynamically classified into charging pile operation data, environmental data, user data, grid load data, and the original multi-dimensional data of the charging station according to the JSON fields.

[0054] S2.2. Use Flume (distributed data collection) to store the charging pile operation data, user data, and grid load data in the distributed relational database. During the storage process, a time series index, user identification index, and geographical location index are established. The original multi-dimensional data of the charging station is partitioned and stored in the distributed file system according to the timestamp and device ID.

[0055] Among them, the distributed relational database uses AES-256 encryption, and the key is managed by the HSM hardware security, and RBAC is used to control data access at different levels.

[0056] S2.3. Use the Presto query engine to achieve efficient SQL query and streaming analysis, support fast retrieval of the composite index, and output the retrieval result as a result set.

[0057] S3. Perform multi-dimensional standardization on the result set, dynamically associate multi-dimensional data using the graph query language, construct a spatio-temporal association graph, extract the relationship features of multi-dimensional data, and generate multi-dimensional feature vectors.

[0058] It includes the following steps:

[0059] S3.1. Based on the retrieval result set, use the Local Outlier Factor algorithm to detect cross-dimensional outliers. For the time series data of charging pile power and grid load, use the Lagrange interpolation method to fill in the missing values. For user data, based on the KNN algorithm, fill in the missing fields (charging duration) according to similar user data.

[0060] S3.2. Use the MinHash (similarity hashing) algorithm to calculate the similarity signatures of multi-dimensional data, detect the similarities between data, and combine with the SimHash (minimal hashing) algorithm. By converting multi-dimensional data into hash values and calculating the similarity of these hash values, streamline data storage and remove redundant information. Use Z-score standardization to map the unified dimension data to the standard normal distribution.

[0061] S3.3. Pre-define the node types as charging pile nodes, user nodes, environment nodes, and grid nodes, and define the edge relationship types as user-charging pile charging relationship, charging pile-environment location relationship, and charging pile-grid influence relationship. Dynamically associate new events through the graph query language. When a new user charging event occurs, dynamically associate environmental data and grid load, generate a spatio-temporal association path, construct a spatio-temporal association graph, and create a combined index of timestamp and geographical hash in the spatio-temporal association graph.

[0062] S3.4. Extract the hotspots of user behavior areas through the spatio-temporal clustering algorithm, extract the coupling intensity characteristics of charging piles and environment-grid, and use the graph attention network to aggregate node features and edge weights to generate multi-dimensional feature vectors.

[0063] Preferably, through cross-dimensional association analysis, intelligent completion, and multi-modal standardization, solve the deficiencies of single-dimensional processing, and provide a high-quality data basis for the global state analysis and dynamic optimization of charging stations. S4. Based on the multi-dimensional feature vectors, combine computer vision to generate a 3D model of the charging station, use the virtual-real mapping equation to dynamically fuse multi-dimensional data into the 3D model, and construct a digital twin model. It includes the following steps:

[0064] S4.1. Generate a 3D model of the charging station through computer vision technology, and dynamically map the operating status of charging piles, user data, and environmental data to the 3D model.

[0065] S4.2. Establish a mathematical mapping relationship between the physical world and the 3D model using the virtual-real mapping equation, and synchronize the operating status of the charging pile, user data, grid load data, and environmental data (serialize and transmit the standardized data through the Protobuf protocol) to the 3D model in real time.

[0066] Among them, the virtual-real mapping equation is expressed as: ;

[0067] In the formula, is time The multi-dimensional data of the charging station in the 3D model at time is time, is expressed as the weighted sum in the process of multi-dimensional data fusion, is the number of multi-dimensional data features, is the weighting coefficient of each data, is time The mapping function of different data, is time The operating status data of the real charging station, is expressed as dynamically adjusting by calculating the topological difference between the multi-dimensional data of the charging station in the 3D model and the multi-dimensional data of the real charging station at each moment, is the spatio-temporal gradient function for calculating the multi-dimensional data of the charging station in the 3D model and the multi-dimensional data of the actual charging station, is the actual multi-dimensional data of the real-time charging station, is the multi-dimensional data of the real-time charging station in the 3D model.

[0068] Among them, The function definition is expressed as: ;

[0069] In the formula, is the adaptive adjustment of the residual error, is to constrain the spatio-temporal continuity of the digital twin model state, is the L2-norm squared error between the real charging station data and the prediction of the charging station in the 3D model.

[0070] Among them, The spatio-temporal smoothing term is expressed as: ;

[0071] In the formula, is the data dimension index, is time at The state component of the charging station in the 3D model, is the L2-norm squared error of the same state component at adjacent time steps, is the previous time.

[0072] S4.3. Dynamically fuse the data of visual modeling and the virtual-real silver snake equation through calculation, and optimize the parameters by combining reinforcement learning to construct a high-precision digital twin model of synchronous physical entities.

[0073] S5. Calculate the residual error between the digital twin model and the real data of the charging station in real time, correct the digital twin model, combine the reinforcement learning algorithm to optimize the interaction efficiency, and dynamically adjust the digital twin model through adaptive gain, which includes the following steps:

[0074] S5.1. Verify the identity by calculating the difference between the data of the digital twin model and the real charging station status data, and use the error feedback to adjust the weights and topological preservation terms in the mapping formula. When the difference exceeds the set synchronous error judgment threshold, the correction mechanism is automatically triggered.

[0075] Encode the charging pile status data output by the virtual-real mapping equation and the environmental data input into the variational autoencoder, and perform anomaly detection through the residual error calculated from the charging pile status data and the environmental data.

[0076] Among them, the calculation of the residual error is expressed as: ;

[0077] In the formula, is the time The difference between the multi-dimensional data of the digital twin model and the multi-dimensional data of the real charging station.

[0078] S5.2. Based on the operation data of the charging station collected continuously for 30 days, statistically analyze the distribution of the residual error of the digital twin model, calculate the mean and standard deviation, and set the synchronous error judgment threshold to .

[0079] Set the synchronous error judgment threshold to ;

[0080] If > , it means that the error correction is triggered and the digital twin model is corrected.

[0081] If ≤ , it means that no correction is required and the digital twin model remains unchanged.

[0082] For the output user data and the real-time grid load data, use the deep deterministic policy gradient algorithm to optimize through reinforcement learning, and optimize the efficiency of virtual-real interaction by dynamically adjusting the constraint factors and weights of the digital twin model.

[0083] S5.3. Input the multi-dimensional data of the charging station using adaptive gain control, and dynamically adjust the gain of the digital twin model through real-time data.

[0084] Among them, the adaptive gain control is expressed as: ;

[0085] In the formula, is the gain adjustment coefficient, is the adjustment factor, is the response function of different data, is the multi-dimensional data input, is the minimum value protection constant.

[0086] Preferably, through computer vision modeling, the three-dimensional geometric information of the charging station is accurately obtained, so that physical objects such as charging piles, power grid equipment, and distribution boxes can be mapped in the digital twin model according to the actual spatial positions, thereby providing an accurate spatial topology basis. On this basis, the virtual-real mapping equation of topology perception introduces a spatio-temporal gradient compensation mechanism, which can adjust the topology structure of the digital twin model in real time according to the changes in the operating state of the charging station, ensuring a high degree of consistency between the digital twin model and the real charging station state. However, during the dynamic mapping process, the operating state and environmental data of the charging pile may be affected by external factors, resulting in the generation of outliers or noise data. An anomaly detection method based on variational autoencoders is adopted to accurately identify anomalies by calculating the residual error between virtual and real data, and trigger automatic correction when the error exceeds the set synchronization error judgment threshold, further improving the reliability of the data and the stability of the model. The adaptive gain control is used to optimize the response of the digital twin model, enabling the digital twin model to continuously optimize according to different operating environments of the charging station, thereby improving the virtual-real interaction efficiency and the overall intelligent level.

[0087] S6. Use a hypergraph neural network to fuse multi-modal data, combine the entropy-regularized multi-objective MDP update strategy and utilize the meta-learning framework to quickly adapt to different tasks, and dynamically optimize the hypergraph topology structure to maximize data association, which includes the following steps:

[0088] S6.1. The neural network is expressed as: ;

[0089] In the formula, is the interaction relationship between devices in the topological graph charging station, is all the multi-dimensional data in the charging station, is the multiple interaction relationship among the charging pile status data, user data, power grid load data, and environmental data, is the weight of all data relationships.

[0090] Use the hypergraph Laplacian matrix to perform regularization and calculate the fused eigenvectors.

[0091] Among them, the hypergraph Laplacian matrix is expressed as: ;

[0092] In the formula, is the fused feature vector, is the regularized hypergraph data, is the feature dimension of the multi-dimensional data, is the adjustment of the linear part output by the bias term, is the Sigmoid function.

[0093] Input the fused feature vector into the multi-objective Markov decision process algorithm to optimize the charging station strategy and balance the relationship between different objectives through entropy regularization.

[0094] S6.2. The entropy-regularized multi-objective MDP update strategy is expressed as: ;

[0095] In the formula, is the total return of the strategy, is the defined time is the strategy for the charging station selection under the environmental data at time is the average value of the calculated return during multiple executions of the strategy, measures the effect of the current strategy action, is the discount factor, represents the influence intensity of the entropy regularization term on the objective function, is expressed as at time, based on the multi-dimensional data of the digital twin model, the entropy of the probability distribution of the possible actions for the strategy selected by the charging station under the environmental data at this moment.

[0096] S6.3. Further utilize the digital twin model to optimize the strategy through meta-learning (the meta-learning strategy outputs the hyperedge weight adjustment instruction to trigger the dynamic update of the hypergraph adjacency matrix), quickly adapt to different tasks and scenarios of the charging station, and use topological graph optimization to calculate the optimal device-to-device interaction strategy.

[0097] Among them, the optimized strategy is expressed as: ;

[0098] In the formula, is the optimized strategy parameter, is the initial strategy parameter, is the set of charging station operation and maintenance environments, is the task index, is the th environment corresponding to the task, is the probability distribution on the charging station operation and maintenance task environment, is the task is the loss function on the task, is the local constraint on the task ​ is the output under the initial policy parameters , is the update step size of the meta learning rate control parameter is the gradient operator

[0099] Among them, the topology graph optimization is expressed as: ;

[0100] In the formula is the optimal topology graph is the topology graph are different data sources in the charging station is the data and the data interaction weight is the data node and the data node association between

[0101] S7. Perform power distribution through a dynamic optimization algorithm, use LSTM-GAN to fuse and predict faults and locate them, and feedback cross-scenario optimization of resource scheduling and load balancing, which includes the following steps:

[0102] S7.1. Use the particle swarm optimization algorithm (PSO), with the core goal of compressing the peak-valley difference of the power grid, and balance the global search and local convergence capabilities through a dynamic weight adjustment strategy. The algorithm inputs include the real-time power demand of charging piles, the power grid load, and user behavior data, and the output is the optimized power distribution instruction for each charging pile. The constraint conditions set the power safety range of the charging pile and user satisfaction.

[0103] S7.2. Real-time collect multi-dimensional data of the charging station (sampling frequency: 1 time / second), construct a data sequence with a time window of 1 hour, and use a double-layer LSTM network (128 hidden units). The input layer receives standardized time series data, and the output layer calculates the fault probability for the next 1 hour through the Sigmoid function. During training, use the cross-entropy loss function and the Adam optimizer, measure the accuracy, the generator receives a random noise vector (dimension 100) and outputs simulated abnormal data; the discriminator is optimized based on the Wasserstein distance (gradient penalty coefficient λ = 10) to enhance the authenticity of the generated data, compare the cosine similarity of the feature vectors of the actual data and the generated abnormal data, and combine the GPS coordinates of the charging pile to output the fault location (such as the radiator of charging pile CZ-001).

[0104] S7.3. Dynamically adjust the proportional coefficient, integral coefficient, and differential coefficient based on reinforcement learning (DDPG algorithm). Through initial value setting, the reward function is the negative reciprocal of the power deviation. Input the target power (from the dynamic optimization algorithm) and the actual power, calculate the deviation, and output the abnormal state.

[0105] S7.4. When in an abnormal state, trigger a shutdown instruction, record the abnormal event, disable the charging function of the faulty device, generate a maintenance work order and push it to the nearest operation and maintenance personnel. The execution record is stored in the Hyperledger Fabric consortium blockchain, using the PBFT consensus algorithm (4 Orderer nodes), and the tampering risk is verified by Monte Carlo attack simulation.

[0106] S7.5. Based on the historical data stored in the blockchain (10 charging station scenarios), train a general policy (MAML algorithm), learn cross-scenario common features (such as the load pattern during the morning rush hour at urban stations), and use the Dijkstra algorithm to calculate the optimal scheduling path for mobile charging piles. The weights are the geographical distance and the road congestion coefficient, and the scheduling response time.

[0107] This embodiment also provides a computer device applicable to the situation of the charging station operation and maintenance method based on a multi-dimensional data system, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the charging station operation and maintenance method based on a multi-dimensional data system proposed in the above embodiment.

[0108] The computer device can be a terminal. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be achieved through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the outer shell of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0109] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for operating and maintaining a charging station based on a multi-dimensional data system as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, abbreviated as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, abbreviated as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, abbreviated as EPROM), programmable read-only memory (Programmable Red-Only Memory, abbreviated as PROM), read-only memory (Read-Only Memory, abbreviated as ROM), magnetic memory, flash memory, magnetic disk or optical disc.

[0110] In summary, through the multi-dimensional data fusion and dynamic optimization mechanism, the present invention significantly improves the accuracy and response efficiency of charging station operation and maintenance. Through the virtual-real mapping equation and the residual error feedback mechanism, the parameters of the digital twin model are dynamically corrected, and the adaptive gain is optimized by combining the reinforcement learning algorithm, reducing the prediction error of the digital twin model. Further, the LSTM-GAN fusion model uses the generative adversarial network to expand abnormal samples, enhancing the recognition ability of rare faults (such as power module attenuation) and improving the positioning accuracy. At the same time, the hypergraph neural network based on the meta-learning framework quickly adapts to different task scenarios, optimizes the power distribution strategy, and realizes the automatic execution of device freezing and power limitation through the blockchain smart contract, ensuring stability under grid load fluctuations and achieving the global optimum of charging station resource scheduling, fault response and grid coordination.

[0111] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for operation and maintenance of a charging station based on a multi-dimensional data system, characterized in that: Including, Collecting multi-dimensional data of the charging station, packaging it into a JSON format data packet and performing preprocessing, where the multi-dimensional data includes charging pile operation data, environmental data, user data, and grid load data; Using the Kafka-Flume combined architecture to parse and classify the JSON format data packet, storing it in the database and establishing a composite index, and using Presto query to perform data joint retrieval and output the retrieval result set; Performing multi-dimensional standardization processing on the retrieval result set, dynamically associating multi-dimensional data using graph query language, constructing a spatio-temporal association graph and extracting multi-dimensional data relationship features to generate multi-dimensional feature vectors; Based on the multi-dimensional feature vectors, combining computer vision to generate a three-dimensional model of the charging station, using the virtual-real mapping equation to dynamically fuse multi-dimensional data into the three-dimensional model to construct a digital twin model; Calculating the residual error between the digital twin model and the real data of the charging station in real time to correct the digital twin model; Combining the reinforcement learning algorithm to optimize the interaction efficiency and dynamically adjusting the digital twin model through adaptive gain, Using the hypergraph neural network to fuse multi-dimensional feature vectors, combining the entropy-regularized multi-objective MDP update strategy and using the meta-learning framework to quickly adapt to different tasks and dynamically optimize the hypergraph topology; Performing power allocation through a dynamic optimization algorithm, using LSTM-GAN to fuse and predict faults and locate them, and feeding back cross-scenario optimization of resource scheduling and load balancing; Using the hypergraph neural network to fuse multi-dimensional feature vectors, combining the entropy-regularized multi-objective MDP update strategy and using the meta-learning framework to quickly adapt to different tasks and dynamically optimize the hypergraph topology includes, Using the hypergraph Laplacian matrix to perform regularization and calculating the fused feature vectors, Inputting the fused feature vectors into the multi-objective Markov decision process algorithm to optimize the charging station strategy and balancing the relationship between different objectives through entropy regularization; The entropy-regularized multi-objective MDP update strategy is expressed as: ; In the formula, The total return of the strategy, is the strategy for charging station selection under the environmental data at the defined time , is the average value of the calculated return during multiple executions of the strategy, is to measure the effect of the current strategic action, is the discount factor, represents the influence intensity of the entropy regularization term on the objective function, is expressed as at time, based on the multi-dimensional data of the digital twin model, the entropy of the probability distribution of the possible actions of the strategy selected by the charging station under the environmental data at this moment; Using the digital twin model to optimize the strategy through meta-learning, quickly adapting to different tasks and scenarios of the charging station, and using topological graph optimization to calculate the optimal device interaction strategy; Among them, the optimization strategy is expressed as: ; In the formula, is the optimized policy parameter, is the initial policy parameter, is the set of charging station operation and maintenance environments, is the task index, is the th environment corresponding to the task, is the probability distribution on the charging station operation and maintenance task environment, is the loss function on the task, is the local constraint on the task, is the output under the initial policy parameter , is the step size for updating the meta learning rate control parameter, is the gradient operator; Among them, the topological graph optimization is expressed as: ; In the formula, is the optimal topology graph, is the topology graph, are different data sources in the charging station, is the data and the data interaction weight, is the data node and the data node association between.

2. The method for charging station operation and maintenance based on a multi-dimensional data system according to claim 1, wherein: The packaging into a JSON format data packet and performing preprocessing includes, Using the Kalman filter algorithm to remove noise, detecting anomalies through the Z-Score method, and using the linear interpolation algorithm for completion.

3. The operation and maintenance method of a charging station based on a multi-dimensional data system according to claim 1, characterized in that: The Kafka-Flume combined architecture for parsing, classifying, storing in the database, and establishing a composite index means, Real-time parsing of the JSON format data packet through Kafka, dynamically classifying it into charging pile operation data, environmental data, user data, grid load data, and the original multi-dimensional data of the charging station based on JSON fields; Using Flume to store the charging pile operation data, user data, and grid load data into a distributed relational database, and establishing a time series index, user identification index, and geographical location index, and storing the charging station in the distributed file system by timestamp and device ID partitioning; 4. The operation and maintenance method of the charging station based on the multi-dimensional data system according to claim 1, characterized in that: The use of graph query language to dynamically associate multi-dimensional data, construct a spatio-temporal association graph, and extract multi-dimensional data relationship features means, Based on the multi-dimensional data of the charging station, the node types are predefined as charging pile nodes, user nodes, environmental nodes, and power grid nodes; the edge relationship types are defined as user-charging pile charging relationship, charging pile-environment location relationship, and charging pile-power grid impact relationship; The graph is updated in real time through the graph query language. When a new user charging event occurs, the environmental data and grid load are dynamically associated to generate a spatio-temporal association path, and a joint index of timestamp and geographical hash is created in the spatio-temporal association graph; The spatio-temporal clustering algorithm is used to extract the hotspots of the user behavior area. Through the calculation of dynamic relationship weights, the coupling strength characteristics of the charging pile and the environment-power grid are extracted to generate a multi-dimensional feature vector.

5. The method for operation and maintenance of a charging station based on a multi-dimensional data system according to claim 1, characterized in that: The use of the virtual-real mapping equation to dynamically fuse multi-dimensional data into a three-dimensional model for constructing a digital twin model means that, By comparing the multi-dimensional data of the charging station collected in real time with the predicted state of the three-dimensional model item by item, the mean absolute error residual is calculated; Using the error residual feedback adjustment, the multi-dimensional data of the charging station collected in real time is compared with the predicted state of the multi-dimensional data of the real-time charging station of the three-dimensional model item by item. The heat dissipation coefficient of the charging pile, the power attenuation factor of the charging pile, and the influence weight of user behavior are calculated, and the mean output error residual is used to perform reverse optimization on the virtual-real mapping equation to construct a digital twin model.

6. The method for charging station operation and maintenance based on a multi-dimensional data system according to claim 1, wherein: The power distribution through the dynamic optimization algorithm includes, Based on the dynamic optimization hypergraph topology, the particle swarm optimization algorithm is adopted. The real-time power demand of the charging pile, the grid load data, and the user behavior data are input, and the optimized power distribution instructions for each charging pile are output.

7. The method for operation and maintenance of a charging station based on a multi-dimensional data system according to claim 1, characterized in that: The use of LSTM-GAN to fuse and predict faults and locate them includes, Based on LSTM-GAN to learn the temporal sequence law of the device state, predict the future fault probability, and generate an adversarial network to simulate abnormal data such as radiator overload and power module attenuation to enhance the training samples; By comparing the characteristic differences between the actual charging station data and the abnormal data, and combining the GPS coordinates of the charging pile to locate the fault, the fault type, location coordinates, and maintenance priority are output.

8. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the charging station operation and maintenance method based on the multi-dimensional data system according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the charging station operation and maintenance method based on the multi-dimensional data system according to any one of claims 1 to 7.

Images