Intelligent dispatching system for substation operation and maintenance vehicles

Abstract

The present invention relates to the technical field of intelligent vehicle dispatching, and specifically to an intelligent dispatching system for substation operation and maintenance vehicles, including a multimodal data integration module, a data fusion analysis module, an adaptive dispatching learning module, an intelligent path planning module, a task management adjustment module, and a safety monitoring interaction module; wherein: the multimodal data integration module is used to collect a variety of data inside and outside the substation in real time; the data fusion analysis module forms an operation and maintenance environment view; the adaptive dispatching learning module is used to learn and optimize vehicle dispatching strategies; the intelligent path planning module calculates the best operation and maintenance vehicle driving route; the task management adjustment module is used to dynamically generate or adjust operation and maintenance tasks. The present invention enhances the response flexibility and safety of substation operation and maintenance by realizing intelligent path planning and task management adjustment, while providing efficient user interaction, significantly improving operation and maintenance efficiency and operational convenience.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle intelligent dispatching, and in particular to an intelligent dispatching system for substation operation and maintenance vehicles. Background Art

[0002] With the rapid development of power systems and the improvement of their intelligence levels, the operation and maintenance management of substations has gradually shown the characteristics of complexity and dynamism. Traditional substation operation and maintenance mainly rely on manual inspections and simple mechanical equipment, which often seem incapable of coping with large-scale, high-frequency operation and maintenance tasks. Especially in emergency situations and complex environmental conditions, traditional methods are difficult to respond to and accurately handle emergencies in real time, which can easily lead to safety risks and low operation and maintenance efficiency. In addition, due to the lack of effective data integration and analysis tools, the problem of information islands within substations is serious, making it difficult to achieve efficient use of data and optimal allocation of resources.

[0003] In the face of these challenges, the present invention aims to solve several key problems existing in the prior art. First, the existing path planning and task scheduling systems lack flexibility and adaptability, and cannot effectively cope with the ever-changing operation and maintenance environment and real-time changing task requirements. Second, the security monitoring and risk management mechanisms are not perfect, especially the rapid response capabilities in emergency situations are insufficient, and potential security risks cannot be effectively predicted and prevented. In addition, the existing systems also have deficiencies in user interaction and feedback processing, and cannot provide real-time and intuitive interactive support, which affects the decision-making and operational efficiency of operation and maintenance personnel.

[0004] Therefore, developing an intelligent operation and maintenance system that integrates advanced data analysis, intelligent scheduling, and efficient safety management has become a necessary measure to improve the efficiency and safety of substation operation and maintenance. Summary of the invention

[0005] Based on the above objectives, the present invention provides an intelligent dispatching system for substation operation and maintenance vehicles.

[0006] The intelligent dispatching system for substation operation and maintenance vehicles includes a multimodal data integration module, a data fusion analysis module, an adaptive dispatching learning module, an intelligent path planning module, a task management adjustment module, and a safety monitoring interaction module; among which:

[0007] Multimodal data integration module: used to collect a variety of data inside and outside the substation in real time, including video surveillance data, sensor monitoring data, meteorological information and real-time feedback from operation and maintenance personnel;

[0008] Data fusion analysis module: receives data from the multimodal data integration module, uses graph neural network algorithm to analyze and process the data, integrates different data, and forms a unified operation and maintenance environment view;

[0009] Adaptive scheduling learning module: Based on the operation and maintenance environment view formed by the data fusion analysis module, the reinforcement learning algorithm is used to continuously learn and optimize the vehicle scheduling strategy. The scheduling plan is automatically adjusted according to the current task requirements, vehicle status and changes in the operation and maintenance environment to achieve dynamic optimization;

[0010] Intelligent Path Planning Module: Works with the Adaptive Scheduling Learning Module to calculate the best route for maintenance vehicles based on the optimized scheduling strategy, specifically based on current real-time traffic information, task priority, and estimated duration, to ensure rapid maintenance response and highest cost efficiency;

[0011] Task management and adjustment module: receives the route arrangement from the intelligent path planning module, and dynamically generates or adjusts the operation and maintenance tasks according to real-time task requirements and emergencies, ensuring the real-time and adaptability of task allocation and optimizing the overall operation and maintenance process;

[0012] Safety monitoring and interaction module: Based on the task allocation of the task management adjustment module, it monitors the safety of vehicles and personnel during the operation and maintenance process, and uses predictive models to identify potential risks. It also provides automatic safety response measures in emergency situations and interacts with operation and maintenance personnel in real time through the user interface.

[0013] Furthermore, the multimodal data integration module includes a video surveillance data acquisition unit, a sensor data acquisition unit, a meteorological information acquisition unit, and an operation and maintenance personnel feedback unit; wherein:

[0014] Video monitoring data acquisition unit: used to collect real-time video data from various areas within the substation. Specifically, through high-definition cameras installed at different predetermined locations within the substation, video images of equipment operation status and on-site environment are obtained 24 hours a day, and the images are compressed and transmitted in real time through video processing technology;

[0015] Sensor data acquisition unit: including temperature sensor, humidity sensor, smoke sensor, vibration sensor, used to monitor the physical parameters of equipment and environment in the substation, and use various sensors to obtain real-time data on temperature, humidity, smoke concentration and vibration;

[0016] Meteorological information collection unit: By connecting to the meteorological data service platform, it can obtain real-time weather information of the substation area, including temperature, humidity, wind speed, and rainfall, to ensure that external environmental factors can be considered when dispatching operation and maintenance vehicles;

[0017] Operation and maintenance personnel feedback unit: Through the intelligent mobile terminal devices equipped by the operation and maintenance personnel, the real-time feedback and status information of the operation and maintenance personnel on site are collected. The operation and maintenance personnel feedback unit includes a mobile application and a real-time communication sub-module. The operation and maintenance personnel can also submit task completion status, equipment abnormality reports and emergency event notifications through mobile terminal devices.

[0018] Furthermore, the data fusion analysis module includes a data preprocessing unit, a graph construction unit, a graph neural network calculation unit and an environment view generation unit; wherein:

[0019] Data preprocessing unit: used to receive data from the multimodal data integration module and perform standardization processing on different types of data, including denoising, normalization and feature extraction, to ensure the consistency and comparability of the data;

[0020] Graph construction unit: used to convert the preprocessed data into a graph structure, where nodes represent various devices or sensors in the substation, and edges represent the connection or association relationship between devices. The overall structure of the substation system is reflected by constructing a graph model;

[0021] Graph neural network computing unit: applies the graph neural network algorithm to the constructed graph structure to perform feature extraction and data fusion. Through multi-layer graph convolution operations, it captures the high-dimensional features of nodes and edges, fuses information from different data, and generates an embedded representation of each node.

[0022] Environmental view generation unit: Integrate the node embedding representation and graph structure information output by the graph neural network computing unit to form a unified operation and maintenance environment view, which is used to display the status, mutual relationships and potential risk points of each equipment in the substation.

[0023] Furthermore, the graph neural network computing unit includes:

[0024] Graph convolution layer subunit: used to apply the graph convolution network algorithm to the constructed graph structure, capturing the high-dimensional features of nodes and edges through multi-layer graph convolution operations. The calculation formula of graph convolution is:

[0025] in, is the node feature matrix of the lth layer, which represents the feature representation of all equipment in the substation at the lth layer; v is the adjacency matrix of the graph, which represents the connection relationship between the devices in the substation; v is the node degree matrix, indicating the degree of each device node; is the weight matrix of the lth layer, which represents the training parameters of the graph convolutional network at the lth layer; σ is the activation function, which is used to introduce nonlinearity; is the input node feature matrix, which represents the original device feature data;

[0026] Feature fusion subunit: It is used to fuse the information of different data. It integrates the multi-dimensional features of each node in the graph convolution layer to achieve feature fusion of different data. Specifically, it updates the embedding representation of each node by aggregating the features of neighboring nodes. The formula is: in, is the embedding representation of node i at the l+1 layer, which represents the feature representation of device i in the substation at the l+1 layer; is the set of neighbor nodes of node i, representing other device nodes directly connected to device i; d i is the degree of node i, indicating the number of connections of device i; d j is the degree of node j, indicating the number of connections of device j; is the weight matrix of the lth layer, which represents the trainable parameters in the lth layer during the feature fusion process; is the feature representation of neighbor node j at layer l, and represents the feature representation of device j in the substation at layer l;

[0027] Node embedding generation subunit: Integrate the node features after multi-layer graph convolution processing to generate the final embedding representation of each node. By superimposing multi-layer graph convolution operations, extract and fuse the features of high-order neighbor nodes layer by layer. The final output node embedding representation can capture the global information and local features in the graph structure.

[0028] Furthermore, the environment view generating unit includes:

[0029] Node feature integration subunit: It is used to receive the node embedding representation output by the graph neural network computing unit, and integrate the node embedding representations at different levels to generate the global feature representation of each node. The specific integration process uses weighted averaging and splicing to operate. The formula is:

[0030] in, is the final feature representation of node i; is the embedding representation of node i at layer l; concat represents the concatenation operation, which concatenates the embedding representations of different layers into a long vector; W c is a weighted matrix used to perform weighted averaging on the concatenated long vectors;

[0031] Graph structure information integration subunit: used to integrate the graph structure information between nodes. Specifically, it uses the edge information output by the graph neural network calculation unit and combines the node features to generate an adjacency matrix that reflects the relationship between nodes. The formula is: A final =D -1 / 2 AD -1 / 2 , where A final is the final adjacency matrix, which indicates the relationship between nodes after integration; D is the degree matrix, which indicates the degree of the node; A is the original adjacency matrix, which indicates the initial node connection relationship;

[0032] Operation and maintenance environment view generation subunit: It is used to combine the final node feature representation generated by the node feature integration unit and the adjacency matrix generated by the graph structure information integration unit to generate a unified operation and maintenance environment view. The operation and maintenance environment view uses visualization technology to display the status, location and relationship of each device in the substation. The generation formula is:

[0033] Among them, V env is the final generated operation and maintenance environment view; f is a visualization function used to map node features and adjacency matrices into intuitive graphical representations.

[0034] Furthermore, the adaptive scheduling learning module includes a state representation unit, an action selection unit, a reward calculation unit and a strategy update unit; wherein:

[0035] State representation unit: It is used to extract the current state information from the operation and maintenance environment view generated by the data fusion analysis module, including task requirements, vehicle status and operation and maintenance environment, and represent the extracted information as the state input of the reinforcement learning algorithm. The formula is: S t ={D t , V t , E t}, where S t is the state vector at time t; D t is the current task demand vector; V t is the current vehicle state vector; E t is the current operation and maintenance environment vector;

[0036] Action selection unit: Apply the policy network in the reinforcement learning algorithm to select the action based on the current state S t Select the optimal scheduling action A t , the action selection unit specifically uses a deep Q network to estimate the Q value of each action, the formula is: A t = argmax a Q(S t ,a;θ), where A t is the scheduling action selected at time t; Q(S t ,a;θ) is the Q value function of the policy network, indicating that in state S t The expected reward for taking action a under θ is the parameter of the policy network.

[0037] Reward calculation unit: used to calculate the reward value R based on the scheduling results t , the reward value is used to reflect the effect of the current scheduling action, the formula is: R t =Y(D t , V t , E t , A t ), where Rt is the reward value at time t; Y is the reward function, which comprehensively considers the task completion, vehicle utilization rate and changes in the operation and maintenance environment;

[0038] Strategy update unit: According to the reward value R t And the experience replay mechanism continuously updates the parameters θ of the policy network to optimize the scheduling strategy. The policy update is specifically achieved by minimizing the loss function, and the formula is:

[0039]

[0040] Among them, α is the learning rate; γ is the discount factor; θ′ is the parameter of the target network; represents the gradient of the parameter θ.

[0041] Furthermore, the intelligent path planning module includes a road condition information acquisition unit, a task priority evaluation unit, a path optimization unit and a path adjustment unit; wherein:

[0042] Road condition information acquisition unit: used to receive and update the road condition information of the substation and its surrounding areas in real time, including traffic congestion, road construction information and emergencies;

[0043] Task priority evaluation unit: Based on the task information provided by the adaptive scheduling learning module, the priority of each task is evaluated. The task priority evaluation takes into account the urgency, impact scope and completion time of the task, and uses weighted scoring for calculation. The formula is: P i =α1E′ i +α2R′ i +α3D′ i , where P i Score the priority of task i; E′ i R' is the urgency score of task i. i Score the impact range of task i; D′ i is the completion time score of task i; α1, α2, α3 are weight parameters;

[0044] Path optimization unit: Based on the road condition information and task priority, the ant colony optimization algorithm is used to calculate the optimal driving route of the operation and maintenance vehicle. The goal of path optimization is to minimize the total driving time and task completion time. The formula is: Among them, T i is the driving time of the vehicle from the current node to task i; W i is the weight of task i, and the task priority P i Proportional to; n is the number of tasks that need to be completed at present; the ant colony optimization algorithm gradually optimizes the driving route by simulating the movement of ants on the path and the update process of pheromones. The formula is: τ ij (t+1)=(1-ρ)τij (t)+Δτ ij , where τ ij (t) represents the pheromone concentration from path i to path j at time t; ρ is the pheromone volatility coefficient; Δτ ij Represents the amount of pheromone newly left on path i to path j within time t;

[0045] Path adjustment unit: During the driving process of the vehicle, the road conditions and task changes are monitored in real time, and the driving route is dynamically adjusted according to the new information to ensure that the vehicle always follows the optimal path. The path adjustment unit uses an incremental path planning algorithm to respond to road condition changes and task updates in a timely manner. The formula is: R new =R old +ΔR, where R new is the new path after adjustment; R old is the original path; ΔR represents the path adjustment increment.

[0046] Further, the task management and adjustment module includes a task monitoring unit, a task generation unit, an emergency response unit, a task adjustment unit and a feedback processing unit; wherein:

[0047] Task monitoring unit: used to monitor the execution status and progress of the current task in real time, and obtain real-time feedback on task execution and vehicle location data by connecting with the operation and maintenance personnel's mobile terminal devices and vehicle location tracking system;

[0048] Task generation unit: Dynamically generates new operation and maintenance tasks based on the route arrangement and current task requirements provided by the intelligent path planning module. The task generation unit comprehensively considers task priority, vehicle availability and current road conditions, generates an adaptive operation and maintenance task list, and assigns it to the corresponding operation and maintenance vehicles;

[0049] Emergency response unit: used to handle emergencies that occur during the operation and maintenance process. Specifically, it receives emergency information from the multimodal data integration unit and the operation and maintenance personnel, evaluates the urgency and scope of the incident, and adjusts the task priority and vehicle scheduling strategy in a timely manner;

[0050] Task adjustment unit: During the task execution process, the task arrangement is dynamically adjusted according to the real-time changing task requirements and emergency information. The task adjustment unit will recalculate the task priority according to the latest task status and environmental information, and optimize the task allocation and vehicle driving route;

[0051] Feedback processing unit: used to collect and process real-time feedback and completion status reports from operation and maintenance personnel during task execution. Through continuous communication with operation and maintenance personnel, it can timely understand problems and suggestions in task execution, and feed back relevant information to the task monitoring unit and task adjustment unit.

[0052] Furthermore, the security monitoring interaction module includes a security monitoring unit, a risk identification unit, a security response unit and a user interaction unit; wherein:

[0053] Safety monitoring unit: used to adjust the task allocation of the task management module and monitor the safety status of vehicles and personnel in real time during the operation and maintenance process. The safety monitoring unit obtains the real-time location, status and environmental information of vehicles and personnel through the GPS positioning system of the operation and maintenance vehicle, on-board sensors and safety equipment worn by the operation and maintenance personnel to ensure the comprehensive collection of safety data;

[0054] Risk identification unit: Uses a preset prediction model to identify potential security risks. By analyzing the data provided by the security monitoring unit, a time series prediction model is used to predict potential risks. The formula is:

[0055] C t =σ(W r ·X t +b r ), where C t is the risk prediction value at time t; W r is the weight matrix of the prediction model; X t is the input feature vector at time t, including vehicle status, personnel location and environmental data; b r is the bias vector; σ is the activation function;

[0056] Safety response unit: used to provide automatic safety response measures in emergency situations when potential risks are detected. The safety response unit includes automatic alarm, emergency shutdown and remote control functions. When the risk identification unit predicts a high risk, the safety response unit will automatically trigger the corresponding function to ensure the safety of operation and maintenance personnel and equipment;

[0057] User Interaction Unit: Provides an intuitive graphical interface for displaying task assignment, security monitoring and risk prediction results. Operation and maintenance personnel can view real-time data, receive alerts and perform operations through the interface.

[0058] Beneficial effects of the present invention:

[0059] The present invention significantly improves the dispatching efficiency and response flexibility of substation operation and maintenance tasks by implementing an intelligent path planning module and a task management and adjustment module. The intelligent path planning module uses an advanced ant colony optimization algorithm, combined with real-time road condition information and task priority, to dynamically calculate the optimal operation and maintenance vehicle driving route, which not only reduces the total driving time and cost of the operation and maintenance tasks, but also improves the processing speed of emergency tasks. At the same time, the task management and adjustment module can dynamically generate or adjust tasks according to real-time changes in on-site conditions, ensuring the efficient execution of operation and maintenance work and the rationality of task allocation.

[0060] The present invention greatly enhances the safety management and risk prevention capabilities in the operation and maintenance process through the combined use of a safety monitoring unit and a risk identification unit. The safety monitoring unit monitors the status of vehicles and personnel in real time to ensure the safety of operation and maintenance personnel, while the risk identification unit effectively predicts potential safety risks through advanced prediction models and provides early warning information in a timely manner. This early warning mechanism makes the response in emergency situations more rapid and accurate, significantly reducing the risk of accidents and possible losses.

[0061] The present invention enables operation and maintenance personnel to monitor the operation and maintenance process more conveniently and make quick decisions through an intuitive graphical interface and real-time data feedback. In addition, the system also supports voice interaction and instant messaging functions, so that operation and maintenance personnel can obtain real-time guidance and feedback when performing tasks, further improving the accuracy and convenience of operation and maintenance task execution. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0063] Figure 1 A schematic diagram of an intelligent dispatching system for operation and maintenance vehicles according to an embodiment of the present invention;

[0064] Figure 2 Schematic diagram of a data fusion analysis module according to an embodiment of the present invention. DETAILED DESCRIPTION

[0065] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0066] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. The words "include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects.

[0067] like Figure 1-Figure 2As shown in the figure, the intelligent dispatching system for substation operation and maintenance vehicles includes a multimodal data integration module, a data fusion analysis module, an adaptive dispatching learning module, an intelligent path planning module, a task management adjustment module and a safety monitoring interaction module; among which:

[0068] Multimodal data integration module: used to collect a variety of data inside and outside the substation in real time, including video surveillance data, sensor monitoring data, meteorological information and real-time feedback from operation and maintenance personnel, providing comprehensive data support for the entire system;

[0069] Data fusion analysis module: receives data from the multimodal data integration module, uses the Graph Neural Networks (GNNs) algorithm to analyze and process the data, integrates different data, and forms a unified operation and maintenance environment view;

[0070] Adaptive scheduling learning module: Based on the operation and maintenance environment view formed by the data fusion analysis module, the reinforcement learning algorithm is used to continuously learn and optimize the vehicle scheduling strategy. The scheduling plan is automatically adjusted according to the current task requirements, vehicle status and changes in the operation and maintenance environment to achieve dynamic optimization;

[0071] Intelligent Path Planning Module: Works with the Adaptive Scheduling Learning Module to calculate the best route for maintenance vehicles based on the optimized scheduling strategy, specifically based on current real-time traffic information, task priority, and estimated duration, to ensure rapid maintenance response and highest cost efficiency;

[0072] Task management and adjustment module: receives the route arrangement from the intelligent path planning module, and dynamically generates or adjusts the operation and maintenance tasks according to real-time task requirements and emergencies, ensuring the real-time and adaptability of task allocation and optimizing the overall operation and maintenance process;

[0073] Safety monitoring and interaction module: Based on the task allocation of the task management adjustment module, it monitors the safety of vehicles and personnel during the operation and maintenance process, and uses predictive models to identify potential risks. It also provides automatic safety response measures in emergency situations and interacts with operation and maintenance personnel in real time through the user interface to ensure that all adjustment measures can be monitored and modified in real time.

[0074] The multimodal data integration module includes a video surveillance data acquisition unit, a sensor data acquisition unit, a meteorological information acquisition unit, and an operation and maintenance personnel feedback unit; among which:

[0075] Video surveillance data acquisition unit: used to collect real-time video data from various areas within the substation. Specifically, through high-definition cameras installed at different predetermined locations within the substation, video images of equipment operation status and on-site environment are obtained 24 hours a day, and the images are compressed and transmitted in real time through video processing technology to ensure efficient transmission and storage of data;

[0076] Sensor data acquisition unit: including temperature sensors, humidity sensors, smoke sensors, and vibration sensors, which are used to monitor the physical parameters of the equipment and environment in the substation. Various sensors are used to obtain data on temperature, humidity, smoke concentration, and vibration in real time, and transmit these data to the central processing system through a wireless sensor network to ensure the real-time and accuracy of the data;

[0077] Meteorological information collection unit: By connecting to the meteorological data service platform, it can obtain real-time weather information of the substation area, including temperature, humidity, wind speed, and rainfall, to ensure that external environmental factors can be considered when dispatching operation and maintenance vehicles;

[0078] Operation and maintenance personnel feedback unit: Through the intelligent mobile terminal devices equipped by the operation and maintenance personnel, the real-time feedback and status information of the operation and maintenance personnel on site are collected. The operation and maintenance personnel feedback unit includes mobile applications and real-time communication sub-modules. The operation and maintenance personnel can also submit task completion status, equipment abnormality reports and emergency event notifications through mobile terminal devices; through the design of the above-mentioned multimodal data integration unit, the system can comprehensively and real-time collect a variety of data inside and outside the substation, providing a solid data foundation for subsequent data analysis and scheduling decisions. The video monitoring data acquisition unit ensures visual monitoring of the equipment operation status, the sensor data acquisition unit provides accurate environmental and equipment physical parameters, and the meteorological information acquisition unit takes into account the impact of the external environment on operation and maintenance work. The operation and maintenance personnel feedback unit ensures timely updating and transmission of on-site information. This real-time collection and integration of multi-source data greatly improves the intelligence level and response speed of the system, and helps to achieve efficient and safe substation operation and maintenance vehicle scheduling.

[0079] The data fusion analysis module includes a data preprocessing unit, a graph construction unit, a graph neural network calculation unit, and an environment view generation unit; wherein:

[0080] Data preprocessing unit: used to receive data from the multimodal data integration module and perform standardization processing on different types of data, including denoising, normalization and feature extraction, to ensure the consistency and comparability of the data;

[0081] Graph construction unit: used to convert the preprocessed data into a graph structure, where nodes represent various devices or sensors in the substation, and edges represent the connection or association relationship between devices. The overall structure of the substation system is reflected by constructing a graph model;

[0082] Graph neural network computing unit: applies the graph neural network algorithm to the constructed graph structure to perform feature extraction and data fusion. Through multi-layer graph convolution operations, it captures the high-dimensional features of nodes and edges, fuses information from different data, and generates an embedded representation of each node.

[0083] Environmental view generation unit: Integrate the node embedding representation and graph structure information output by the graph neural network calculation unit to form a unified operation and maintenance environment view, which is used to display the status, mutual relationships and potential risk points of each device in the substation, and provide intuitive data support for subsequent scheduling decisions; through the design of the above-mentioned data fusion and analysis module, the system can make full use of the powerful analysis capabilities of the graph neural network to deeply fuse and comprehensively analyze multi-source data. The data preprocessing unit ensures the consistency and quality of the input data. The graph construction unit effectively represents the complex relationship of the substation system through the graph structure model. The graph neural network calculation unit deeply explores the characteristics of nodes and edges, and the environmental view generation unit displays this information intuitively. Finally, the unified operation and maintenance environment view formed by this module provides a scientific basis and support for the intelligent scheduling of substation operation and maintenance vehicles, and improves the overall intelligence level and decision-making accuracy of the system.

[0084] The graph neural network computing unit includes:

[0085] Graph convolution layer subunit: used to apply the graph convolution network algorithm to the constructed graph structure, capturing the high-dimensional features of nodes and edges through multi-layer graph convolution operations. The calculation formula of graph convolution is:

[0086] in, is the node feature matrix of the lth layer, which represents the feature representation of all equipment in the substation at the lth layer; v is the adjacency matrix of the graph, which represents the connection relationship between the devices in the substation; v is the node degree matrix, indicating the degree of each device node; is the weight matrix of the lth layer, which represents the training parameters of the graph convolutional network at the lth layer; σ is the activation function, and the ReLU activation function is commonly used to introduce nonlinearity; is the input node feature matrix, which represents the original device feature data;

[0087] Feature fusion subunit: It is used to fuse the information of different data. It integrates the multi-dimensional features of each node in the graph convolution layer to achieve feature fusion of different data. Specifically, it updates the embedding representation of each node by aggregating the features of neighboring nodes. The formula is: in, is the embedding representation of node i at the l+1 layer, which represents the feature representation of device i in the substation at the l+1 layer; is the set of neighbor nodes of node i, representing other device nodes directly connected to device i; d i is the degree of node i, indicating the number of connections of device i; d j is the degree of node j, indicating the number of connections of device j; is the weight matrix of the lth layer, which represents the trainable parameters in the lth layer during the feature fusion process; is the feature representation of neighbor node j at layer l, and represents the feature representation of device j in the substation at layer l;

[0088] Node embedding generation subunit: Integrate the node features after multi-layer graph convolution processing to generate the final embedding representation of each node. By superimposing multi-layer graph convolution operations, extract and fuse the features of high-order neighbor nodes layer by layer. The final output node embedding representation can capture the global information and local features in the graph structure. Through the design of the above-mentioned graph neural network computing unit, the system can efficiently perform in-depth analysis on the complex graph structure of the substation. The graph convolution layer unit captures the high-dimensional features of nodes and edges through multi-layer graph convolution operations. The feature fusion unit effectively fuses the multi-dimensional feature information from different data. The node embedding generation unit finally generates a high-quality embedding representation for each node. In this way, the system can not only accurately represent the status and relationship of each node in the substation, but also provide rich information to support subsequent scheduling decisions, greatly improving the intelligence level of the system and the accuracy of decision-making.

[0089] The environment view generation unit includes:

[0090] Node feature integration subunit: It is used to receive the node embedding representation output by the graph neural network computing unit, and integrate the node embedding representations at different levels to generate the global feature representation of each node. The specific integration process uses weighted averaging and splicing to operate. The formula is:

[0091] in, is the final feature representation of node i; is the embedding representation of node i at layer l; concat represents the concatenation operation, which concatenates the embedding representations of different layers into a long vector; W c is a weighted matrix used to perform weighted averaging on the concatenated long vectors;

[0092] Graph structure information integration subunit: used to integrate the graph structure information between nodes. Specifically, it uses the edge information output by the graph neural network calculation unit and combines the node features to generate an adjacency matrix that reflects the relationship between nodes. The formula is: A final =D -1 / 2 AD -1 / 2 , where A final is the final adjacency matrix, which indicates the relationship between nodes after integration; D is the degree matrix, which indicates the degree of the node; A is the original adjacency matrix, which indicates the initial node connection relationship;

[0093] Operation and maintenance environment view generation subunit: It is used to combine the final node feature representation generated by the node feature integration unit and the adjacency matrix generated by the graph structure information integration unit to generate a unified operation and maintenance environment view. The operation and maintenance environment view uses visualization technology to display the status, location and relationship of each device in the substation. The generation formula is:

[0094] Among them, V env is the final generated operation and maintenance environment view; f is a visualization function used to map node features and adjacency matrices into intuitive graphical representations; through the design of the above-mentioned environment view generation unit, the system can effectively integrate the node embedding representation and graph structure information output by the graph neural network calculation unit to form a unified operation and maintenance environment view. The node feature integration unit ensures that the global feature representation of each node is accurate, and the graph structure information integration unit effectively reflects the relationship between nodes. The operation and maintenance environment view generation unit visualizes this information, so that the operation and maintenance personnel can intuitively understand the status and relationship of each equipment in the substation. This comprehensive environment view provides comprehensive and real-time data support for intelligent scheduling decisions, greatly improving the intelligence level of the system and the accuracy of decision-making.

[0095] The adaptive scheduling learning module includes a state representation unit, an action selection unit, a reward calculation unit, and a strategy update unit; wherein:

[0096] State representation unit: It is used to extract the current state information from the operation and maintenance environment view generated by the data fusion analysis module, including task requirements, vehicle status and operation and maintenance environment, and represent the extracted information as the state input of the reinforcement learning algorithm. The formula is: S t ={D t , V t , E t}, where S t is the state vector at time t; D t is the current task demand vector; V t is the current vehicle state vector; E t is the current operation and maintenance environment vector;

[0097] Action selection unit: Apply the policy network in the reinforcement learning algorithm to select the action based on the current state S t Select the optimal scheduling action A t , the action selection unit specifically uses the deep Q network (DQN) to estimate the Q value of each action, the formula is: A t = argmax a Q(S t ,a;θ), where A t is the scheduling action selected at time t; Q(S t,a;θ) is the Q value function of the policy network, indicating that in state S t The expected reward for taking action a under θ is the parameter of the policy network.

[0098] Reward calculation unit: used to calculate the reward value R based on the scheduling results t , the reward value is used to reflect the effect of the current scheduling action, the formula is: R t =Y(D t , V t , E t , A t ), where R t is the reward value at time t; Y is the reward function, which comprehensively considers the task completion, vehicle utilization rate and changes in the operation and maintenance environment;

[0099] Strategy update unit: According to the reward value R t And the experience replay mechanism continuously updates the parameters θ of the policy network to optimize the scheduling strategy. The policy update is specifically achieved by minimizing the loss function, and the formula is:

[0100]

[0101] Among them, α is the learning rate; γ is the discount factor; θ′ is the parameter of the target network; Represents the gradient of parameter θ; through the design of the above-mentioned adaptive scheduling learning module, the system can apply the reinforcement learning algorithm to continuously learn and optimize the vehicle scheduling strategy based on the operation and maintenance environment view generated by the data fusion analysis module. The state representation unit ensures the accurate information input of the current task requirements, vehicle status and operation and maintenance environment. The action selection unit selects the optimal scheduling action through the strategy network. The reward calculation unit calculates the reward value according to the scheduling result. The strategy update unit continuously optimizes the scheduling strategy through the experience playback mechanism. Finally, the module realizes automatic adjustment of the scheduling plan according to the actual operation and maintenance needs and environmental changes, improves the intelligence level and scheduling efficiency of the system, and ensures the efficiency and safety of substation operation and maintenance.

[0102] The intelligent path planning module includes a road condition information acquisition unit, a task priority evaluation unit, a path optimization unit, and a path adjustment unit; wherein:

[0103] Traffic information acquisition unit: used to receive and update the traffic information of the substation and its surrounding areas in real time, including traffic congestion, road construction information and emergencies. It is received by connecting to the traffic information service platform and the substation internal monitoring system to ensure the accuracy and timeliness of the traffic data required for route planning.

[0104] Task priority evaluation unit: Based on the task information provided by the adaptive scheduling learning module, the priority of each task is evaluated. The task priority evaluation takes into account the urgency, impact scope and completion time of the task, and uses weighted scoring for calculation. The formula is: P i =α1E′ i +α2R′ i +α3D′ i , where P i Score the priority of task i; E′ i R' is the urgency score of task i. i Score the impact range of task i; D′ i is the completion time score of task i; α1, α2, α3 are weight parameters;

[0105] Path optimization unit: Based on the road condition information and task priority, the ant colony optimization algorithm is used to calculate the optimal driving route of the operation and maintenance vehicle. The goal of path optimization is to minimize the total driving time and task completion time. The formula is: Among them, T i is the driving time of the vehicle from the current node to task i; W i is the weight of task i, and the task priority P i Proportional to; n is the number of tasks that need to be completed at present; the ant colony optimization algorithm gradually optimizes the driving route by simulating the movement of ants on the path and the update process of pheromones. The formula is: τ ij (t+1)=(1-ρ)τ ij (t)+Δτ ij , where τ ij (t) represents the pheromone concentration from path i to path j at time t; ρ is the pheromone volatility coefficient; Δτ ij Represents the amount of pheromone newly left on path i to path j within time t;

[0106] Path adjustment unit: During the driving process of the vehicle, the road conditions and task changes are monitored in real time, and the driving route is dynamically adjusted according to the new information to ensure that the vehicle always follows the optimal path. The path adjustment unit uses an incremental path planning algorithm to respond to road condition changes and task updates in a timely manner. The formula is: R new =R old +ΔR, where R new is the new path after adjustment; R oldis the original path; ΔR represents the path adjustment increment, which is calculated based on the latest road condition information and task requirements; through the design of the above-mentioned intelligent path planning module, the system can dynamically calculate the best operation and maintenance vehicle driving route according to the current real-time road condition information, task priority and expected duration. The road condition information acquisition unit ensures the accuracy of the data required for path planning, the task priority evaluation unit reasonably evaluates the priority of each task, the path optimization unit uses the ant colony optimization algorithm to calculate the optimal driving route, and the path adjustment unit dynamically adjusts the route during the vehicle driving process to ensure the continuous effectiveness of the optimal path. In this way, the system not only improves the efficiency of operation and maintenance vehicle scheduling, but also greatly reduces the task completion time and driving cost, and improves the overall efficiency and response speed of substation operation and maintenance.

[0107] The task management and adjustment module includes a task monitoring unit, a task generation unit, an emergency response unit, a task adjustment unit and a feedback processing unit; wherein:

[0108] Task monitoring unit: used to monitor the execution status and progress of the current task in real time, and obtain real-time feedback on task execution and vehicle location data by connecting with the operation and maintenance personnel's mobile terminal devices and vehicle location tracking system;

[0109] Task generation unit: Dynamically generates new operation and maintenance tasks based on the route arrangement and current task requirements provided by the intelligent path planning module. The task generation unit comprehensively considers task priority, vehicle availability and current road conditions, generates an adaptive operation and maintenance task list, and assigns it to the corresponding operation and maintenance vehicles;

[0110] Emergency response unit: used to handle emergencies that occur during the operation and maintenance process. Specifically, it receives emergency information from the multimodal data integration unit and operation and maintenance personnel, evaluates the urgency and scope of the incident, and timely adjusts task priorities and vehicle scheduling strategies to ensure that emergencies are quickly responded to and handled;

[0111] Task adjustment unit: During the task execution process, the task arrangement is dynamically adjusted according to the real-time changing task requirements and emergency information. The task adjustment unit will recalculate the task priority according to the latest task status and environmental information, optimize task allocation and vehicle driving routes, and ensure the efficient and orderly operation and maintenance work;

[0112] Feedback processing unit: used to collect and process real-time feedback and completion status reports from operation and maintenance personnel during task execution. Through continuous communication with operation and maintenance personnel, the system can promptly understand problems and suggestions in task execution, and feed back relevant information to the task monitoring unit and task adjustment unit for necessary adjustments and optimizations. Through the design of the above-mentioned task management adjustment module, the system can dynamically generate or adjust operation and maintenance tasks according to real-time task requirements and emergencies after receiving the route arrangement of the intelligent path planning module. The task monitoring unit ensures real-time grasp of task execution, the task generation unit generates an adaptive operation and maintenance task list based on the latest information, the emergency response unit quickly responds to and handles emergencies, the task adjustment unit dynamically adjusts task arrangements according to changes, and the feedback processing unit promptly collects and processes real-time feedback from operation and maintenance personnel. This design not only improves the flexibility and response speed of operation and maintenance tasks, but also ensures the efficient and orderly execution of tasks, and improves the overall efficiency and safety of substation operation and maintenance.

[0113] The security monitoring interaction module includes a security monitoring unit, a risk identification unit, a security response unit and a user interaction unit; wherein:

[0114] Safety monitoring unit: used to adjust the task allocation of the task management module and monitor the safety status of vehicles and personnel in real time during the operation and maintenance process. The safety monitoring unit obtains the real-time location, status and environmental information of vehicles and personnel through the GPS positioning system of the operation and maintenance vehicle, on-board sensors and safety equipment worn by the operation and maintenance personnel to ensure the comprehensive collection of safety data;

[0115] Risk identification unit: Uses a preset prediction model to identify potential security risks. By analyzing the data provided by the security monitoring unit, a time series prediction model is used to predict potential risks. The formula is:

[0116] C t =σ(W r ·X t +b r ), where C t is the risk prediction value at time t; W r is the weight matrix of the prediction model; X t is the input feature vector at time t, including vehicle status, personnel location and environmental data; b r is the bias vector; σ is the activation function, including sigmoid or ReLU activation function;

[0117] Safety response unit: used to provide automatic safety response measures in emergency situations when potential risks are detected. The safety response unit includes automatic alarm, emergency shutdown and remote control functions. When the risk identification unit predicts a high risk, the safety response unit will automatically trigger the corresponding function to ensure the safety of operation and maintenance personnel and equipment;

[0118] User interaction unit: provides an intuitive graphical interface for displaying task allocation, safety monitoring and risk prediction results. Operation and maintenance personnel can view real-time data, receive alarms and perform operations through the interface. The user interaction unit also supports voice interaction and instant messaging functions, which facilitates operation and maintenance personnel to provide timely feedback and adjust work plans. Through the further design of the above-mentioned task management adjustment module, the system can monitor the safety of vehicles and personnel during the operation and maintenance process based on task allocation, identify potential risks by adopting prediction models, and provide automatic safety response measures in emergency situations. The safety monitoring unit ensures comprehensive safety data collection, the risk identification unit effectively predicts potential risks through a time series prediction model, and the safety response unit automatically triggers safety measures when the risk is high. The user interaction unit interacts with the operation and maintenance personnel in real time through an intuitive interface and instant messaging functions. In this way, the system not only improves the safety and reliability of operation and maintenance tasks, but also enhances the adaptability and work efficiency of operation and maintenance personnel, ensuring the efficient, safe and orderly operation and maintenance of substations.

[0119] The present invention is intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. Intelligent dispatching system for substation operation and maintenance vehicles, characterized by: It includes multimodal data integration module, data fusion analysis module, adaptive scheduling learning module, intelligent path planning module, task management adjustment module and security monitoring interaction module; among which: Multimodal data integration module: used to collect a variety of data inside and outside the substation in real time, including video surveillance data, sensor monitoring data, meteorological information and real-time feedback from operation and maintenance personnel; Data fusion analysis module: receives data from the multimodal data integration module, uses graph neural network algorithm to analyze and process the data, integrates different data, and forms a unified operation and maintenance environment view; Adaptive scheduling learning module: Based on the operation and maintenance environment view formed by the data fusion analysis module, the reinforcement learning algorithm is used to continuously learn and optimize the vehicle scheduling strategy. The scheduling plan is automatically adjusted according to the current task requirements, vehicle status and changes in the operation and maintenance environment to achieve dynamic optimization; The adaptive scheduling learning module includes a state representation unit, an action selection unit, a reward calculation unit and a strategy update unit; wherein: State representation unit: It is used to extract the current state information from the operation and maintenance environment view generated by the data fusion analysis module, including task requirements, vehicle status and operation and maintenance environment, and represent the extracted information as the state input of the reinforcement learning algorithm. The formula is: S t ={D t , V t , E t }, where S t is the state vector at time t; D t is the current task demand vector; V t is the current vehicle state vector; E t is the current operation and maintenance environment vector; Action selection unit: Apply the policy network in the reinforcement learning algorithm to select the action based on the current state S t Select the optimal scheduling action A t , the action selection unit specifically uses a deep Q network to estimate the Q value of each action, the formula is: A t = argmax a Q(S t , a; θ), where A t is the scheduling action selected at time t; Q(S t , a; θ) is the Q value function of the policy network, indicating that in state S t The expected reward for taking action a under θ is the parameter of the policy network. Reward calculation unit: used to calculate the reward value R based on the scheduling results t , the reward value is used to reflect the effect of the current scheduling action, the formula is: R t =Y(D t , V t , E t , A t ), where R t is the reward value at time t; Y is the reward function, which comprehensively considers the task completion, vehicle utilization and changes in the operation and maintenance environment; Strategy update unit: According to the reward value R t And the experience replay mechanism continuously updates the parameters θ of the policy network to optimize the scheduling strategy. The policy update is specifically achieved by minimizing the loss function, and the formula is: Among them, α is the learning rate; γ is the discount factor; θ′ is the parameter of the target network; Represents the gradient of the parameter θ; Intelligent Path Planning Module: Works with the Adaptive Scheduling Learning Module to calculate the best route for maintenance vehicles based on the optimized scheduling strategy, specifically based on current real-time traffic information, task priority, and estimated duration, to ensure rapid maintenance response and highest cost efficiency; The intelligent path planning module includes a road condition information acquisition unit, a task priority evaluation unit, a path optimization unit and a path adjustment unit; wherein: Road condition information acquisition unit: used to receive and update the road condition information of the substation and its surrounding areas in real time, including traffic congestion, road construction information and emergencies; Task priority evaluation unit: Based on the task information provided by the adaptive scheduling learning module, the priority of each task is evaluated. The task priority evaluation takes into account the urgency, impact scope and completion time of the task, and uses weighted scoring for calculation. The formula is: P i =α1E′ i +α2R′ i +α3D′ i , where P i Score the priority of task i; E′ i R' is the urgency score of task i. i Score the impact range of task i; D′ i is the completion time score of task i; α1, α2, α3 are weight parameters; Path optimization unit: Based on the road condition information and task priority, the ant colony optimization algorithm is used to calculate the optimal driving route of the operation and maintenance vehicle. The goal of path optimization is to minimize the total driving time and task completion time. The formula is: Among them, T i is the driving time of the vehicle from the current node to task i; W i is the weight of task i, and the task priority P i Proportional to; n is the number of tasks that need to be completed at present; the ant colony optimization algorithm gradually optimizes the driving route by simulating the movement of ants on the path and the update process of pheromones. The formula is: τ ij (t+1)=(1-ρ)τ ij (t)+Δτ ij , where τ ij (t) represents the pheromone concentration from path i to path j at time t; ρ is the pheromone volatility coefficient; Δτ ij Represents the amount of pheromone newly left on path i to path j within time t; Path adjustment unit: During the driving process of the vehicle, the road conditions and task changes are monitored in real time, and the driving route is dynamically adjusted according to the new information to ensure that the vehicle always follows the optimal path. The path adjustment unit uses an incremental path planning algorithm to respond to road condition changes and task updates in a timely manner. The formula is: R new =R old +ΔR, where R new is the new path after adjustment; R old is the original path; ΔR represents the path adjustment increment; Task management and adjustment module: receives the route arrangement from the intelligent path planning module, and dynamically generates or adjusts the operation and maintenance tasks according to real-time task requirements and emergencies, ensuring the real-time and adaptability of task allocation and optimizing the overall operation and maintenance process; Safety monitoring and interaction module: Based on the task allocation of the task management adjustment module, it monitors the safety of vehicles and personnel during the operation and maintenance process, and uses predictive models to identify potential risks. It also provides automatic safety response measures in emergencies and interacts with operation and maintenance personnel in real time through the user interface.

2. The intelligent dispatching system for substation operation and maintenance vehicles according to claim 1 is characterized in that: The multimodal data integration module includes a video surveillance data acquisition unit, a sensor data acquisition unit, a meteorological information acquisition unit, and an operation and maintenance personnel feedback unit; wherein: Video monitoring data acquisition unit: used to collect real-time video data from various areas within the substation. Specifically, through high-definition cameras installed at different predetermined locations within the substation, video images of equipment operation status and on-site environment are obtained 24 hours a day, and the images are compressed and transmitted in real time through video processing technology; Sensor data acquisition unit: including temperature sensor, humidity sensor, smoke sensor, vibration sensor, used to monitor the physical parameters of equipment and environment in the substation, and use various sensors to obtain real-time data on temperature, humidity, smoke concentration and vibration; Meteorological information collection unit: By connecting to the meteorological data service platform, it can obtain real-time weather information of the substation area, including temperature, humidity, wind speed, and rainfall, to ensure that external environmental factors can be considered when dispatching operation and maintenance vehicles; Operation and maintenance personnel feedback unit: Through the intelligent mobile terminal devices equipped by the operation and maintenance personnel, the real-time feedback and status information of the operation and maintenance personnel on site are collected. The operation and maintenance personnel feedback unit includes a mobile application and a real-time communication sub-module. The operation and maintenance personnel can also submit task completion status, equipment abnormality reports and emergency event notifications through mobile terminal devices.

3. The intelligent dispatching system for substation operation and maintenance vehicles according to claim 1 is characterized in that: The data fusion analysis module includes a data preprocessing unit, a graph construction unit, a graph neural network calculation unit and an environment view generation unit; wherein: Data preprocessing unit: used to receive data from the multimodal data integration module and perform standardization processing on different types of data, including denoising, normalization and feature extraction, to ensure the consistency and comparability of the data; Graph construction unit: used to convert the preprocessed data into a graph structure, where nodes represent various devices or sensors in the substation, and edges represent the connection or association relationship between devices. The overall structure of the substation system is reflected by constructing a graph model; Graph neural network computing unit: applies the graph neural network algorithm to the constructed graph structure to perform feature extraction and data fusion. Through multi-layer graph convolution operations, it captures the high-dimensional features of nodes and edges, fuses information from different data, and generates an embedded representation of each node. Environmental view generation unit: Integrate the node embedding representation and graph structure information output by the graph neural network computing unit to form a unified operation and maintenance environment view, which is used to display the status, mutual relationships and potential risk points of each equipment in the substation.

4. The intelligent dispatching system for substation operation and maintenance vehicles according to claim 3 is characterized in that: The graph neural network computing unit includes: Graph convolution layer subunit: used to apply the graph convolution network algorithm to the constructed graph structure, capturing the high-dimensional features of nodes and edges through multi-layer graph convolution operations. The calculation formula of graph convolution is: in, is the node feature matrix of the lth layer, which represents the feature representation of all equipment in the substation at the lth layer; v is the adjacency matrix of the graph, which represents the connection relationship between the devices in the substation; v is the node degree matrix, indicating the degree of each device node; is the weight matrix of the lth layer, which represents the training parameters of the graph convolutional network at the lth layer; σ is the activation function, which is used to introduce nonlinearity; is the input node feature matrix, which represents the original device feature data; Feature fusion subunit: It is used to fuse the information of different data. It integrates the multi-dimensional features of each node in the graph convolution layer to achieve feature fusion of different data. Specifically, it updates the embedding representation of each node by aggregating the features of neighboring nodes. The formula is: in, is the embedding representation of node i at the l+1 layer, which represents the feature representation of device i in the substation at the l+1 layer; is the set of neighbor nodes of node i, representing other device nodes directly connected to device i; d i is the degree of node i, indicating the number of connections of device i; d j is the degree of node j, indicating the number of connections of device j; is the weight matrix of the lth layer, which represents the trainable parameters in the lth layer during the feature fusion process; is the feature representation of neighbor node j at layer l, and represents the feature representation of device j in the substation at layer l; Node embedding generation subunit: Integrate the node features after multi-layer graph convolution processing to generate the final embedding representation of each node. By superimposing multi-layer graph convolution operations, extract and fuse the features of high-order neighbor nodes layer by layer. The final output node embedding representation can capture the global information and local features in the graph structure.

5. The intelligent dispatching system for substation operation and maintenance vehicles according to claim 4 is characterized in that: The environment view generating unit comprises: Node feature integration subunit: It is used to receive the node embedding representation output by the graph neural network computing unit, and integrate the node embedding representations at different levels to generate the global feature representation of each node. The specific integration process uses weighted averaging and splicing to operate. The formula is: in, is the final feature representation of node i; is the embedding representation of node i at layer l; concat represents the concatenation operation, which concatenates the embedding representations of different layers into a long vector; W c is a weighted matrix used to perform weighted averaging on the concatenated long vectors; Graph structure information integration subunit: used to integrate the graph structure information between nodes. Specifically, it uses the edge information output by the graph neural network calculation unit and combines the node features to generate an adjacency matrix that reflects the relationship between nodes. The formula is: A final =D -1 / 2 AD -1 / 2 , where A final is the final adjacency matrix, which indicates the relationship between nodes after integration; D is the degree matrix, which indicates the degree of the node; A is the original adjacency matrix, which indicates the initial node connection relationship; Operation and maintenance environment view generation subunit: It is used to combine the final node feature representation generated by the node feature integration unit and the adjacency matrix generated by the graph structure information integration unit to generate a unified operation and maintenance environment view. The operation and maintenance environment view uses visualization technology to display the status, location and relationship of each device in the substation. The generation formula is: Among them, V env is the final generated operation and maintenance environment view; f is a visualization function used to map node features and adjacency matrices into intuitive graphical representations.

6. The intelligent dispatching system for substation operation and maintenance vehicles according to claim 1 is characterized in that: The task management and adjustment module includes a task monitoring unit, a task generation unit, an emergency response unit, a task adjustment unit and a feedback processing unit; wherein: Task monitoring unit: used to monitor the execution status and progress of the current task in real time, and obtain real-time feedback on task execution and vehicle location data by connecting with the operation and maintenance personnel's mobile terminal devices and vehicle location tracking system; Task generation unit: Dynamically generates new operation and maintenance tasks based on the route arrangement and current task requirements provided by the intelligent path planning module. The task generation unit comprehensively considers task priority, vehicle availability and current road conditions, generates an adaptive operation and maintenance task list, and assigns it to the corresponding operation and maintenance vehicles; Emergency response unit: used to handle emergencies that occur during the operation and maintenance process. Specifically, it receives emergency information from the multimodal data integration unit and the operation and maintenance personnel, evaluates the urgency and scope of the incident, and adjusts the task priority and vehicle scheduling strategy in a timely manner; Task adjustment unit: During the task execution process, the task arrangement is dynamically adjusted according to the real-time changing task requirements and emergency information. The task adjustment unit will recalculate the task priority according to the latest task status and environmental information, and optimize the task allocation and vehicle driving route; Feedback processing unit: used to collect and process real-time feedback and completion status reports from operation and maintenance personnel during task execution. Through continuous communication with operation and maintenance personnel, it can timely understand problems and suggestions in task execution, and feed back relevant information to the task monitoring unit and task adjustment unit.

7. The intelligent dispatching system for substation operation and maintenance vehicles according to claim 1 is characterized in that: The security monitoring interaction module includes a security monitoring unit, a risk identification unit, a security response unit and a user interaction unit; wherein: Safety monitoring unit: used to adjust the task allocation of the task management module and monitor the safety status of vehicles and personnel in real time during the operation and maintenance process. The safety monitoring unit obtains the real-time location, status and environmental information of vehicles and personnel through the GPS positioning system of the operation and maintenance vehicle, on-board sensors and safety equipment worn by the operation and maintenance personnel to ensure the comprehensive collection of safety data; Risk identification unit: uses preset prediction models to identify potential security risks, analyzes data provided by the security monitoring unit, and uses time series prediction models to predict potential risks; Safety response unit: used to provide automatic safety response measures in emergency situations when potential risks are detected. The safety response unit includes automatic alarm, emergency shutdown and remote control functions. When the risk identification unit predicts a high risk, the safety response unit will automatically trigger the corresponding function to ensure the safety of operation and maintenance personnel and equipment; User Interaction Unit: Provides an intuitive graphical interface for displaying task assignment, security monitoring and risk prediction results. Operation and maintenance personnel can view real-time data, receive alerts and perform operations through the interface.

8. The intelligent dispatching system for substation operation and maintenance vehicles according to claim 7 is characterized in that: The formula of the time series prediction model in the risk identification unit is: t =σ(W r ·X t +b r ), where C t is the risk prediction value at time t; W r is the weight matrix of the prediction model; X t is the input feature vector at time t, including vehicle status, personnel location and environmental data; b r is the bias vector; σ is the activation function.

Images