Distribution Network Operation and Maintenance Testing Method and Device Based on Micro Modular Processing

Abstract

The present invention discloses a distribution network operation and maintenance testing method and device based on micro modular processing, which relates to the technical field of distribution network operation and maintenance testing. Through the comprehensive monitoring coverage of the distribution network system by distributed monitoring, the present invention enables the monitoring tasks to be distributed to different micro modules, improving the comprehensiveness and timeliness of monitoring and making the decision-making more flexible. By adopting an intelligent decision-making model based on deep reinforcement learning, the system can autonomously learn and optimize the decision-making strategy according to the real-time system state and environmental changes, improving the flexibility and adaptability of the system. In addition, faults can be processed in real time. The reasoning module adopts a lightweight processing model and runs quickly on edge devices, improving the real-time response ability and reliability of the system, making the operation and maintenance management of the distribution network more efficient and reliable.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network operation and maintenance testing, and specifically to a distribution network operation and maintenance testing method and device based on micro modular processing. Background Art

[0002] Distribution network operation and maintenance testing is usually carried out for the operation and maintenance of the distribution network to ensure the reliability, safety and efficiency of the distribution system, and to detect and eliminate potential problems in a timely manner, so as to ensure the stability and continuity of power supply. Distribution network operation and maintenance testing usually includes regular inspections of various equipment in the distribution system, such as transformers, switchgear, and distribution panels, measuring and recording electrical parameters, such as voltage, current, and power factor, evaluating the electrical performance of the system, testing the reliability and response performance of protection devices, and checking parameters such as the grounding resistance and grounding polarity of the grounding system.

[0003] For example, the patent with publication number CN116093806A discloses a distribution network equipment operation and maintenance method and system. The method includes obtaining the operation and maintenance data of several distribution network equipment in a distribution area, and calculating key operation and maintenance analysis indicators such as the external force damage failure rate, hidden danger rectification rate, lightning strike failure rate, ground flash density, user failure rate, and user sectionalizing switch action rate according to the operation and maintenance data; establishing an operation and maintenance strategy correlation index analysis model according to the key operation and maintenance analysis indicators, and calculating the external force damage evaluation factor of each distribution network equipment through the operation and maintenance strategy correlation index analysis model; adjusting the operation and maintenance strategies of each distribution network equipment according to the external force damage evaluation factor of each distribution network equipment, and controlling each distribution network equipment to perform operation and maintenance according to the adjusted operation and maintenance strategies; calculating the total evaluation factor according to the proportion of various operation and maintenance failures and the external force damage evaluation factor, and evaluating the operation and maintenance quality of the distribution area.

[0004] Although the above solution has the above advantages, traditional distribution network operation and maintenance testing methods are limited to specific areas or specific types of monitoring tasks, unable to achieve full coverage of the distribution network system, resulting in blind spots and loopholes in the system, and easily missing potential faults and abnormal situations. At the same time, the decision-making mode of traditional methods is relatively rigid and cannot be flexibly adjusted according to the real-time system state and environmental changes, with poor adaptability. Therefore, there is an urgent need for a distribution network operation and maintenance testing method and device based on micro modular processing to solve such problems. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention provides a distribution network operation and maintenance testing method and device based on micro modular processing, which solves the problems existing in the prior art, such as being limited to specific areas or specific types of monitoring tasks, unable to achieve full coverage of the distribution network system, and the decision-making mode of traditional methods being relatively rigid and unable to be flexibly adjusted according to the real-time system state and environmental changes.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0007] The present invention provides a distribution network operation and maintenance test method based on micro modular processing, including:

[0008] Step 1. Conduct distributed real-time monitoring. The distributed monitoring module distributes the monitoring tasks to different micro modules, and each micro module is responsible for monitoring a specific area to provide comprehensive coverage of the distribution network system. The monitoring process uses distributed message passing for data exchange and synchronization between modules;

[0009] Step 2. Heterogeneous data fusion. Based on data from different types of sensors and monitoring devices, including current, voltage, and temperature, the heterogeneous data fusion is used to integrate and fuse multi-source data, construct a data model for the distribution network system, and then process and filter the data;

[0010] The method for constructing the data model of the distribution network system is as follows:

[0011] Collect data from different types of sensors and monitoring devices, including current I, voltage V, and temperature T, from each node of the distribution network system;

[0012] Denoise, fill in missing values, and standardize the original data. Extract features from the data collected by each sensor to obtain high-dimensional feature vectors, and then use an autoencoder to reduce the dimension of the high-dimensional features to obtain low-dimensional feature representations;

[0013] Fuse the low-dimensional feature vectors from different sensors to construct an overall feature representation of the distribution network system;

[0014] Use the fused feature vectors to construct a data model for the distribution network system;

[0015] Based on the deep reinforcement learning method, construct an intelligent decision-making model for the distribution network system;

[0016] Step 3. Real-time data processing and inference. Edge computing devices are deployed at the edge nodes of the distribution network system, and edge computing is used for real-time data processing and inference. Here, the real-time inference model is a lightweight processing model. At the same time, for faults and abnormal conditions in the distribution network system, real-time inference rules are deployed to quickly diagnose and handle the faults;

[0017] Step 4. Monitoring of operating status and data changes. Based on the distributed monitoring results, the operating status and data changes of the distribution network system are monitored in real time, including voltage fluctuations and current anomalies. Based on the heterogeneous data fusion and real-time inference results, real-time diagnosis and early warning of system faults and anomalies are carried out;

[0018] At the same time, a visualization interface is provided to display the real-time monitoring data and fault diagnosis results.

[0019] The present invention is further configured such that in the distributed monitoring step, first, the monitoring area is divided, and the task volume of each monitoring area is estimated:

[0020] The monitoring area of the distribution network system is divided into N sub-areas, and each sub-area corresponds to a micro-module M i (i = 1, 2,..., N), where i represents the number of the module, and for each sub-area M i Estimate the monitoring task volume T i , T i represents the monitoring task volume of the i-th micro-module, and the task volume T i is estimated according to the regional power demand and line density factors;

[0021] Establish a task allocation optimization model to minimize the system monitoring task volume, and at the same time evenly allocate the monitoring tasks of each micro-module;

[0022] The optimization objective function is to minimize the weighted sum of the monitoring task volumes of each micro-module, representing the minimization of the overall system monitoring task volume, then: where wi represents the weight of each sub-area M i and at the same time establish constraint conditions;

[0023] The constraint conditions are: the monitoring task volume of each micro-module M i cannot exceed its maximum bearable task volume, that is, Ti ≤ Ci, where Ci represents the maximum task volume of each micro-module M i ;

[0024] Each monitoring task is only assigned to one micro-module, that is Tj ∩ Tk = φ, where j and k both represent the indexes of the micro-modules, that is, two different micro-modules, and Tj ∩ Tk = φ means that the intersection of the monitoring task sets Tj and Tk is an empty set, that is, the task sets Tj and Tk have no common tasks;

[0025] Then use a heuristic algorithm to solve the task allocation optimization model to obtain a task allocation scheme;

[0026] According to the optimization result, dynamically allocate the monitoring tasks to each micro-module;

[0027] Adopt a task scheduling algorithm and perform data exchange and synchronization between modules through a distributed message passing mechanism;

[0028] The present invention is further configured such that the steps of constructing an intelligent decision-making model for the distribution network system include:

[0029] Model the distribution network system as a reinforcement learning environment, including an intelligent agent. The intelligent agent performs different actions in different states and obtains corresponding rewards;

[0030] Let the state represent the current state of the distribution network system, t represent the current moment, the action at represent the action taken by the agent in the state st, and the reward rt represent the feedback obtained by the agent after executing the action at in the state st;

[0031] Use a deep neural network as the policy network of the agent, take the state as the input, and output the probability distribution of the action;

[0032] Let the policy network of the agent be represented as πθ(as) = softmax(fθ(s)), where fθ(s) represents the network parameters, πθ(as) represents the probability of the agent taking the action a in the state s, θ represents the parameters of the policy network, and softmax represents the activation function;

[0033] The present invention is further set as: The steps of constructing the intelligent decision-making model of the distribution network system further include:

[0034] Construct a reward function rt to evaluate the goodness or badness of the agent taking different actions at in different states st;

[0035] Use deep Q-learning to optimize the policy network of the agent. By interacting with the environment, the agent adjusts its policy according to the reward signal and gradually learns the optimized distribution network decision-making strategy. The Q-value update method is: Where α represents the learning rate, γ represents the discount factor, st+1 represents the next state transferred to by the agent after executing the action at, and a' is the optimal action taken in the next state st+1;

[0036] The present invention is further set as: In step 3, the real-time inference rule deployment method is:

[0037] Select an autoencoder, an isolation forest, and a deep neural network to construct an anomaly detection model;

[0038] Deploy a loss function to measure the difference between the original data and the reconstructed data;

[0039] Then, based on the output result of the anomaly detection model, deploy real-time inference rules, set a threshold, and when the anomaly score output by the anomaly detection model, trigger the real-time inference rules;

[0040] According to the anomaly type, device information, and historical data, select the corresponding processing plan and trigger the corresponding maintenance operation;

[0041] The present invention also provides a distribution network operation and maintenance test device based on micro modular processing, including:

[0042] A real-time inference system, which is responsible for real-time monitoring of the operation state of the distribution network system, quickly diagnosing and handling system failures and abnormal conditions;

[0043] The real-time inference system includes a distributed monitoring module and a data processing module;

[0044] The operation status monitoring and early warning module monitors the operation status and data changes of the distribution network system in real time;

[0045] The distributed monitoring module is responsible for distributing monitoring tasks to different micro-modules to achieve comprehensive coverage of the distribution network system;

[0046] The distributed monitoring module includes:

[0047] The task division unit divides the monitoring area into sub-areas and estimates the monitoring task volume of each sub-area;

[0048] The allocation optimization unit establishes a task allocation optimization model to provide balanced task allocation;

[0049] The task dynamic unit dynamically allocates monitoring tasks according to the optimization results and realizes data exchange and synchronization between modules through a message passing mechanism;

[0050] The data processing module is responsible for fusing and processing data from different sensors and constructing a data model of the distribution network system;

[0051] The data processing module includes:

[0052] The extraction and dimensionality reduction unit extracts features from each sensor data, reduces the dimensionality and fuses them into an overall feature representation, and constructs a data model of the distribution network system based on the fused feature vector;

[0053] The present invention is further configured as a distribution network operation and maintenance test device based on micro modular processing, and further includes:

[0054] The intelligent decision-making module is used to construct an intelligent decision-making model for the distribution network system to realize intelligent management and optimization of the distribution network system;

[0055] The intelligent decision-making module models the distribution network system as a reinforcement learning environment, sets states, actions and rewards, and uses a deep neural network to construct a policy network for the intelligent agent to realize the mapping from states to actions;

[0056] Then, the deep Q-learning algorithm is used to optimize the policy network, and gradually learn the optimized distribution network decision-making strategy;

[0057] The present invention also discloses an electronic device, including: a processor and a memory connected to the processor; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory to implement the distribution network operation and maintenance test method based on micro modular processing;

[0058] The present invention also discloses a computer-readable storage medium, in which computer-executable instructions are stored, and when the computer-executable instructions are executed by a processor, they are used to implement a distribution network operation and maintenance test method based on micro modular processing;

[0059] The present invention also discloses a computer program product, including a computer program, which when executed by a processor, implements a distribution network operation and maintenance test method based on micro modular processing.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] In the present invention, the distribution network operation and maintenance test device is modularized. First, the distributed monitoring module distributes the monitoring tasks to different micro modules, and each micro module is responsible for monitoring a specific area, realizing comprehensive coverage of the distribution network system. Through the steps of task division, task assignment optimization, and task dynamic assignment, the balanced distribution and efficient execution of the monitoring tasks are ensured, making the monitoring work of the distribution network system more comprehensive and effective;

[0062] Secondly, the data processing module is responsible for fusing and processing the data from different sensors, constructing a data model of the distribution network system, preprocessing the original data, feature extraction, and dimensionality reduction fusion, providing data support for intelligent decision-making. Then, the intelligent decision-making module models the distribution network system as a reinforcement learning environment, constructs a policy network of the intelligent agent using a deep neural network, and through interaction with the environment, gradually learns an optimized distribution network decision-making strategy, enabling the distribution network system to autonomously learn and optimize the decision-making strategy according to different environments and requirements, improving the intelligence level and adaptability of the system;

[0063] In the present invention, the real-time inference method adopts a lightweight processing model, which runs quickly on edge devices, realizes rapid diagnosis and processing of faults and abnormal conditions in the distribution network system, deploys real-time inference rules to achieve rapid diagnosis and processing of faults, and improves the reliability and stability of the system;

[0064] In summary, through the comprehensive monitoring coverage of the distribution network system by distributed monitoring, the present invention enables the monitoring tasks to be distributed to different micro modules, improves the comprehensiveness and timeliness of monitoring, is more flexible in decision-making, adopts an intelligent decision-making model based on deep reinforcement learning, enables the system to autonomously learn and optimize the decision-making strategy according to the real-time system state and environmental changes, improves the flexibility and adaptability of the system. In addition, faults can be processed in real time, and the inference module adopts a lightweight processing model, which runs quickly on edge devices, improves the real-time response ability and reliability of the system, making the operation and maintenance management of the distribution network more efficient and reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1Flowchart of the distribution network operation and maintenance test method based on micro modular processing according to the present invention;

[0066] Figure 2 Architecture diagram of the distribution network operation and maintenance test device based on micro modular processing according to the present invention. Detailed implementation manners

[0067] In order to enable those skilled in the art of the present technology to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0068] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0069] The present invention will be further described in detail below with reference to the accompanying drawings:

[0070] Embodiment

[0071] Please refer to Figure 1 - Figure 2 , the present invention provides a distribution network operation and maintenance test method based on micro modular processing, including:

[0072] Step 1. Perform distributed real-time monitoring. The distributed monitoring module distributes the monitoring tasks to different micro modules, and each micro module is responsible for monitoring a specific area to provide comprehensive coverage of the distribution network system. The monitoring process uses distributed message passing for data exchange and synchronization between modules;

[0073] In the distributed monitoring step, first divide the monitoring area and estimate the task volume of each monitoring area:

[0074] Divide the monitoring area of the distribution network system into N sub-areas, and each sub-area corresponds to a micro module M i (i = 1, 2,..., N), i represents the i-th module, and for each sub-area Mi Estimate the monitoring task volume T i , T i represents the monitoring task volume of the i-th micro-module, and the task volume T i is estimated according to the regional power demand and line density factors;

[0075] Establish a task allocation optimization model to minimize the system monitoring task volume, and at the same time evenly distribute the monitoring tasks of each micro-module to ensure comprehensive monitoring coverage of the distribution network system;

[0076] The optimization objective function is to minimize the weighted sum of the monitoring task volumes of each micro-module, representing the minimization of the overall system monitoring task volume, then: where wi represents each sub-region M i of the weight, and at the same time establish constraints;

[0077] The constraints are: the monitoring task volume of each micro-module M i cannot exceed its maximum bearable task volume, that is, Ti ≤ Ci, where Ci represents each micro-module M i of the maximum task volume;

[0078] Each monitoring task is only assigned to one micro-module, that is Tj ∩ Tk = φ, where j and k both represent the indices of the micro-modules, that is, two different micro-modules, and Tj ∩ Tk = φ means that the intersection of the monitoring task sets Tj and Tk is an empty set, that is, the task sets Tj and Tk have no common tasks;

[0079] Then use the heuristic algorithm to solve the task allocation optimization model to obtain the task allocation scheme;

[0080] According to the optimization results, dynamically allocate the monitoring tasks to each micro-module to achieve balanced allocation and efficient execution of the monitoring tasks;

[0081] Adopt a task scheduling algorithm to ensure that the monitoring tasks of each micro-module can be completed in a timely manner, and perform data exchange and synchronization between modules through a distributed message passing mechanism;

[0082] Step 2. Heterogeneous data fusion. Based on the data of different types of sensors and monitoring devices, including current, voltage, and temperature, use heterogeneous data fusion to integrate and fuse multi-source data, construct a data model for the distribution network system, and then process and filter the data;

[0083] The construction method of the distribution network system data model is:

[0084] Collect data from different types of sensors and monitoring devices, including current I, voltage V, and temperature T, from each node of the distribution network system;

[0085] Denoise the original data, fill in the missing values, and perform data standardization. Extract features from the data collected by each sensor to obtain high-dimensional feature vectors, and then use an autoencoder to reduce the dimension of the high-dimensional features to obtain low-dimensional feature representations;

[0086] Fuse the low-dimensional feature vectors from different sensors to construct an overall feature representation of the distribution network system;

[0087] Use the fused feature vectors to construct a data model of the distribution network system;

[0088] Based on the deep reinforcement learning method, construct an intelligent decision-making model for the distribution network system;

[0089] The steps to construct the intelligent decision-making model for the distribution network system include:

[0090] Model the distribution network system as a reinforcement learning environment, including an agent. The agent performs different actions in different states and obtains corresponding rewards;

[0091] Let the state represent the current state of the distribution network system, t represent the current time, the action at represent the action taken by the agent in the state st, and the reward rt represent the feedback obtained by the agent after executing the action at in the state st;

[0092] Use a deep neural network as the policy network of the agent, take the state as the input, and output the probability distribution of the action;

[0093] Let the policy network of the agent be represented as πθ(as) = softmax(fθ(s)), where fθ(s) represents the network parameters, πθ(as) represents the probability of the agent taking the action a in the state s, θ represents the parameters of the policy network, and softmax represents the activation function;

[0094] The steps to construct the intelligent decision-making model for the distribution network system also include:

[0095] Construct a reward function rt to evaluate the quality of the agent taking different actions at in different states st:

[0096] The reward function is used to guide the agent to learn the strategy to optimize the performance of the distribution network system;

[0097] Use deep Q-learning to optimize the policy network of the agent. Through interaction with the environment, the agent adjusts the policy according to the reward signal and gradually learns the optimized distribution network decision-making strategy. The Q-value update method: Among them, α represents the learning rate, γ represents the discount factor, st+1 represents the next state transferred to by the agent after executing the action at, and a' is the optimal action taken in the next state st+1;

[0098] Build an intelligent decision-making model for the distribution network system based on deep reinforcement learning, and use the reinforcement learning algorithm to optimize the model so that it can adapt to different distribution network scenarios;

[0099] Step 3. Real-time data processing and inference. Edge computing devices are deployed at the edge nodes of the distribution network system. Edge computing is used for real-time data processing and inference. Here, the real-time inference model is a lightweight processing model that can run quickly on edge devices. At the same time, for faults and abnormal situations in the distribution network system, real-time inference rules are deployed to quickly diagnose and handle faults;

[0100] In Step 3, the deployment method of the real-time inference rules is as follows:

[0101] Select an autoencoder, an isolation forest, and a deep neural network to build an anomaly detection model;

[0102] Deploy a loss function to measure the difference between the original data and the reconstructed data;

[0103] Then, based on the output results of the anomaly detection model, deploy real-time inference rules and set a threshold. When the anomaly score output by the anomaly detection model is reached, the real-time inference rules are triggered;

[0104] Select the corresponding processing scheme according to the anomaly type, device information, and historical data, and trigger the corresponding maintenance operation;

[0105] Step 4. Monitoring of the operating state and data changes. Based on the distributed monitoring results, the operating state and data changes of the distribution network system are monitored in real time, including voltage fluctuations and current anomalies. Based on heterogeneous data fusion and real-time inference results, real-time diagnosis and early warning of system faults and anomalies are carried out;

[0106] At the same time, a visualization interface is provided to display real-time monitoring data and fault diagnosis results to help operation and maintenance personnel discover and handle problems in a timely manner.

[0107] The present invention also provides a distribution network operation and maintenance test device based on micro modular processing. The distribution network operation and maintenance test device in this application aims to achieve real-time monitoring, data fusion, intelligent decision-making, and fault handling of the distribution network. It adopts a modular design and includes:

[0108] A real-time inference system responsible for real-time monitoring of the operating state of the distribution network system and quickly diagnosing and handling system faults and abnormal situations;

[0109] The real-time inference system includes a distributed monitoring module and a data processing module;

[0110] An operating state monitoring and early warning module for real-time monitoring of the operating state and data changes of the distribution network system;

[0111] The distributed monitoring module is responsible for distributing monitoring tasks to different micro-modules to achieve comprehensive coverage of the distribution network system;

[0112] The distributed monitoring module includes:

[0113] The task division unit divides the monitoring area into sub-areas and estimates the monitoring task volume of each sub-area;

[0114] The allocation optimization unit establishes a task allocation optimization model to provide balanced task allocation;

[0115] The task dynamic unit dynamically distributes monitoring tasks according to the optimization results and realizes data exchange and synchronization between modules through the message passing mechanism;

[0116] The data processing module is responsible for fusing and processing data from different sensors and constructing a data model of the distribution network system;

[0117] The data processing module includes:

[0118] The extraction and dimensionality reduction unit extracts features from each sensor data, reduces the dimensionality and fuses them into an overall feature representation, and constructs a data model of the distribution network system based on the fused feature vector;

[0119] The intelligent decision-making module is used to construct an intelligent decision-making model for the distribution network system for intelligent management and optimization of the distribution network system;

[0120] The intelligent decision-making module models the distribution network system as a reinforcement learning environment, sets states, actions and rewards, and uses a deep neural network to construct a policy network for the agent to realize the mapping from state to action;

[0121] Then, the deep Q-learning algorithm is used to optimize the policy network, and gradually learn the optimized distribution network decision-making strategy.

[0122] In the present invention, the distribution network operation and maintenance test device is modularized. First, the distributed monitoring module distributes monitoring tasks to different micro-modules, and each micro-module is responsible for monitoring a specific area to achieve comprehensive coverage of the distribution network system. Through the steps of task division, task allocation optimization and task dynamic allocation, the balanced allocation and efficient execution of monitoring tasks are ensured, making the monitoring work of the distribution network system more comprehensive and effective;

[0123] Secondly, the data processing module is responsible for fusing and processing data from different sensors, constructing a data model for the distribution network system, preprocessing the raw data, feature extraction, and dimensionality reduction fusion, providing data support for intelligent decision-making. Then, the intelligent decision-making module models the distribution network system as a reinforcement learning environment, constructs a policy network for the intelligent agent using a deep neural network, and through interaction with the environment, gradually learns an optimized distribution network decision-making strategy, enabling the distribution network system to autonomously learn and optimize decision-making strategies according to different environments and requirements, improving the intelligence level and adaptability of the system;

[0124] In the present invention, the real-time inference method adopts a lightweight processing model, which runs quickly on edge devices to achieve rapid diagnosis and processing of faults and abnormal conditions in the distribution network system. Deploying real-time inference rules realizes rapid diagnosis and processing of faults, improving the reliability and stability of the system;

[0125] In summary, the distributed monitoring of the present invention provides comprehensive monitoring coverage for the distribution network system, enabling monitoring tasks to be distributed to different micro-modules, improving the comprehensiveness and timeliness of monitoring, and being more flexible in decision-making. Using an intelligent decision-making model based on deep reinforcement learning enables the system to autonomously learn and optimize decision-making strategies according to real-time system states and environmental changes, improving the flexibility and adaptability of the system. In addition, faults can be processed in real time, and the inference module adopts a lightweight processing model, which runs quickly on edge devices, improving the real-time response ability and reliability of the system, making the operation and maintenance management of the distribution network more efficient and reliable.

[0126] The above content is only to illustrate the technical idea of the present invention and cannot be used to limit the protection scope of the present invention. Any modification made on the basis of the technical solution according to the technical idea proposed by the present invention falls within the protection scope of the claims of the present invention.

Claims

1. A distribution network operation and maintenance test method based on micro-modular processing, characterized in that: include: Step 1. Perform distributed real-time monitoring. The distributed monitoring module distributes the monitoring tasks to different micro-modules. Each micro-module is responsible for monitoring a specific area, providing comprehensive coverage of the distribution network system. The monitoring process uses distributed message transmission to exchange and synchronize data between modules. In the distributed monitoring step, the monitoring area is first divided and the task volume of each monitoring area is estimated: The monitoring area of ​​the distribution network system is divided into N sub-areas, each of which corresponds to a micromodule M. i (i=1,2,...,N), i represents the module number, for each sub-area M i Estimated monitoring task volume T i ,T i represents the monitoring task volume of the i-th micromodule, and the task volume T i Estimate based on regional power demand and line density factors; Establish a task allocation optimization model to minimize the system monitoring task volume and evenly distribute the monitoring tasks of each micro-module; The optimization objective function is to minimize the weighted sum of the monitoring tasks of each micro-module, which means minimizing the overall monitoring task volume of the system. Then: Among them, wi represents each sub-region M i The weights of the , and at the same time establish constraints; The constraints are: each micromodule M i The monitoring task volume cannot exceed the maximum task volume it can carry, that is, Ti≤Ci, where Ci represents each micromodule M i The maximum amount of tasks; Each monitoring task is assigned to only one micromodule, i.e. Tj∩Tk=φ, where j and k both represent the index of micromodules, i.e. two different micromodules, and Tj∩Tk=φ means that the intersection of monitoring task sets Tj and Tk is an empty set, i.e. task sets Tj and Tk have no common tasks; Then, the heuristic algorithm is used to solve the task allocation optimization model and obtain the task allocation plan; Based on the optimization results, the monitoring tasks are dynamically assigned to each micromodule; It uses task scheduling algorithm and distributed message passing mechanism to exchange and synchronize data between modules; Step 2. Heterogeneous data fusion: Based on the data from different types of sensors and monitoring equipment, including current, voltage, and temperature, heterogeneous data fusion is used to integrate and fuse multi-source data, build a distribution network system data model, and then process and filter the data; The distribution network system data model is constructed as follows: Collect data from different types of sensors and monitoring equipment from each node of the distribution network system, including current I, voltage V, and temperature T; De-noise the raw data, fill in missing values, and perform data standardization. Perform feature extraction on the data collected by each sensor to obtain a high-dimensional feature vector. Then use the autoencoder to reduce the dimension of the high-dimensional features to obtain a low-dimensional feature representation. The low-dimensional feature vectors from different sensors are fused to construct the overall feature representation of the distribution network system; Use the fused feature vectors to build a data model for the distribution network system; Based on deep reinforcement learning methods, an intelligent decision-making model for distribution network system is constructed; Step 3. Real-time data processing and reasoning: Edge computing devices are deployed at the edge nodes of the distribution network system, and edge computing is used for real-time data processing and reasoning. The real-time reasoning model here is a lightweight processing model. At the same time, for distribution network system faults and abnormal conditions, real-time reasoning rules are deployed to quickly diagnose and handle faults; Step 4. Operation status and data change monitoring: Based on the distributed monitoring results, the operation status and data changes of the distribution network system are monitored in real time, including voltage fluctuations and current anomalies. Based on heterogeneous data fusion and real-time reasoning results, real-time diagnosis and early warning of system faults and anomalies are carried out; It also provides a visual interface to display real-time monitoring data and fault diagnosis results.

2. The distribution network operation and maintenance test method based on micro modular processing according to claim 1 is characterized in that: The steps to build an intelligent decision-making model for the distribution network system include: The distribution network system is modeled as a reinforcement learning environment, including an agent, which performs different actions in different states and obtains corresponding rewards; Let state represent the current state of the distribution network system, t represent the current time, action at represent the action taken by the agent in state st, and reward rt represent the feedback obtained by the agent st after executing action at; Use a deep neural network as the agent’s policy network, taking the state as input and outputting the probability distribution of the action; Assume that the agent’s policy network is represented as πθ(as)=softmax(fθ(s)), where fθ(s) represents the network parameters, πθ(as) represents the probability that the agent takes action a in state s, θ represents the parameters of the policy network, and softmax represents the activation function.

3. The distribution network operation and maintenance test method based on micro modular processing according to claim 2 is characterized in that: The steps of building the intelligent decision-making model of the distribution network system also include: Construct a reward function rt to evaluate the effectiveness of the agent in taking different actions at in different states st; Use deep Q learning to optimize the agent's policy network. By interacting with the environment, the agent adjusts its strategy based on the reward signal and gradually learns the optimized distribution network decision strategy. The Q value is updated in the following way: Among them, α represents the learning rate, γ represents the discount factor, st+1 represents the next state that the agent transfers to after executing action at, and a' is the optimal action taken in the next state st+1.

4. The distribution network operation and maintenance test method based on micro modular processing according to claim 3 is characterized in that: In step 3, the real-time inference rules are deployed as follows: Use autoencoders, isolation forests, and deep neural networks to build anomaly detection models; The deployment loss function measures the difference between the original data and the reconstructed data; Then, based on the output results of the anomaly detection model, deploy real-time inference rules and set thresholds. When the anomaly detection model outputs an abnormal score, the real-time inference rules are triggered; Select the corresponding processing plan based on the exception type, equipment information and historical data, and trigger the corresponding maintenance operation.

5. A distribution network operation and maintenance test device based on micro-modular processing, adopting a distribution network operation and maintenance test method based on micro-modular processing as claimed in any one of claims 1 to 4, characterized in that: include: Real-time reasoning system, responsible for real-time monitoring of the operating status of the distribution network system, and rapid diagnosis and processing of system failures and abnormal conditions; The real-time reasoning system includes a distributed monitoring module and a data processing module; Operation status monitoring and early warning module, real-time monitoring of the operation status and data changes of the distribution network system; Distributed monitoring module, responsible for distributing monitoring tasks to different micro-modules, to provide comprehensive coverage of the distribution network system; Distributed monitoring modules include: The task division unit divides the monitoring area into sub-areas and estimates the monitoring task volume of each sub-area; Allocation optimization unit, establishes task allocation optimization model and provides balanced task allocation; Task dynamic unit, dynamically allocates monitoring tasks according to optimization results, and realizes data exchange and synchronization between modules through message passing mechanism; The data processing module is responsible for fusing and processing data from different sensors and building a data model for the distribution network system; The data processing module includes: Extract dimensionality reduction units, extract features from each sensor data, reduce the dimension and fuse them into an overall feature representation, and build a data model of the distribution network system based on the fused feature vector.

6. The distribution network operation and maintenance test device based on micro modular processing according to claim 5, characterized in that: The distribution network operation and maintenance test device based on micro-modular processing also includes: Intelligent decision-making module, used to build an intelligent decision-making model for the distribution network system, and to intelligently manage and optimize the distribution network system; The intelligent decision-making module models the distribution network system as a reinforcement learning environment, sets the state, action and reward, uses a deep neural network to build the agent's strategy network, and realizes the mapping from state to action; Then, the deep Q-learning algorithm is used to optimize the policy network and gradually learn the optimized distribution network decision-making strategy.

7. An electronic device, characterized in that: include: A processor, and a memory connected to the processor; The memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 4.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 4 when executed by a processor.

9. A computer program product, characterized in that The invention comprises a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 4.

Images