Intelligent power distribution management method and system based on big data

Abstract

The invention discloses an intelligent power distribution management method and system based on big data, and relates to the technical field of intelligent management of a power system, comprising a data acquisition module, a communication network construction module, a data preprocessing module, a connection relationship among a load balancing module and an intelligent decision module and a distribution condition of the connection relationship in a power distribution system. Deep mining and real-time transmission of multi-source heterogeneous data are realized through a distributed data acquisition unit and an edge computing chip; dynamically adjusting a load distribution strategy by using a nonlinear optimization algorithm to realize global load balancing; and an operation state evaluation report and fault early warning are generated through the deep learning model. The problems that an existing system is difficult to comprehensively analyze multi-source data and lacks a dynamic load balancing mechanism can be effectively solved, and the intelligent management level of a power distribution system is improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent power system management technology, specifically to an intelligent power distribution management method and system based on big data. Background Technology

[0002] With the rapid development of smart grids and big data technologies, big data-based intelligent distribution management methods and systems have demonstrated significant advantages in improving distribution efficiency, optimizing resource allocation, and achieving accurate fault early warning. As a crucial component of modern power grids, the core objective of intelligent distribution management systems is to enhance system reliability and operational efficiency through real-time monitoring and analysis of the distribution system's operating status. However, existing intelligent distribution management systems and technical solutions still have many shortcomings. Current technologies primarily rely on linear reconstruction of historical data and analysis of local load characteristics, lacking the ability to deeply mine and comprehensively analyze multi-source heterogeneous big data, making it difficult to fully reflect the dynamic operating status of the distribution system. Furthermore, existing technologies have relatively simple compensation mechanisms for load balancing, failing to fully consider nonlinear relationships in complex scenarios, which may lead to decreased early warning accuracy and thus affect the overall system performance. These problems limit the effectiveness of existing technologies in practical applications, especially when facing diverse and dynamic distribution demands. Existing methods struggle to meet the needs of modern smart grids for refined management and intelligent decision support. Therefore, there is an urgent need for a solution that can achieve refined management and intelligent decision support throughout the entire lifecycle of the distribution system through deep learning and big data analytics, in order to improve system reliability, operational efficiency, and adaptability to complex scenarios. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a smart power distribution management method and system based on big data, which solves the problems of existing power distribution management systems being unable to perform in-depth mining and comprehensive analysis of multi-source heterogeneous data, and lacking a dynamic load balancing compensation mechanism for complex scenarios.

[0004] To achieve the above objectives, the present invention provides an intelligent power distribution management system based on big data, comprising: Data acquisition module: Distributed data acquisition units are set up at each key node of the power distribution system, and the nodes are divided into multiple area groups according to their physical location and electrical topology. Each area group is equipped with an area aggregation node to receive the data within that group. Communication network construction module: An edge computing chip and a wireless communication module are embedded in each distributed data acquisition unit. The acquisition unit is uniquely numbered and a data identifier is generated based on the data acquisition timestamp and node number. Low-latency communication links between adjacent acquisition units are established sequentially from the power distribution master station to the regional aggregation node. Data preprocessing module: The distributed data acquisition units within the regional group send the collected real-time data to the regional aggregation node. The regional aggregation node performs preliminary cleaning and compression on the received data, and then packages and transmits it to the nearest backbone communication node. Load balancing module: The backbone communication node dynamically monitors the load data within and across regions, and adjusts the load distribution strategy of each region based on historical load characteristics and real-time data through a nonlinear optimization algorithm to achieve global load balancing; Intelligent decision-making module: The power distribution master station receives data from the backbone communication nodes, performs comprehensive analysis of the data through a deep learning model to generate an operational status assessment report of the power distribution system, and provides accurate early warnings of potential fault points.

[0005] Preferably, an edge computing chip and a wireless communication module are embedded in each distributed data acquisition unit, and the acquisition unit is uniquely numbered, including the following steps: An edge computing chip and a wireless communication module are embedded in each distributed data acquisition unit. The edge computing chip is used to perform preliminary processing and anomaly detection of local data. A suitable data storage format is selected for efficient storage and subsequent analysis. The wireless communication module selects different communication technologies according to the environment in which the node is located. In high-bandwidth demand scenarios, a 5G module is used; in low-power wide-area scenarios, an NB-IoT module is used; and in short-range communication scenarios, a Zigbee module is used. Assign a unique device identifier to each distributed data acquisition unit and create a power distribution system topology map to record the physical location, electrical connections, and number of shortest path nodes between each acquisition unit and the power distribution master station.

[0006] Preferably, establishing low-latency communication links between adjacent acquisition units sequentially from the power distribution master station to the regional aggregation node includes the following steps: From the main power distribution station to the regional aggregation node, low-latency communication links between adjacent acquisition units are established sequentially in the power distribution network through wireless communication modules according to the power distribution system topology. Regional aggregation nodes establish efficient data transmission links with the nearest backbone communication node via the MQTT-SN protocol to ensure the reliability and real-time performance of data transmission.

[0007] Preferably, the power distribution master station receives data from the backbone communication node, performs comprehensive analysis of the data using a deep learning model to generate an operational status assessment report for the power distribution system, and provides precise early warnings for potential fault points, including the following steps: The power distribution master station aggregates the data transmitted by the backbone communication nodes and uploads it to the cloud big data platform; the cloud big data platform performs standardized processing on the data to establish real-time operation data tables and historical load data tables for the power distribution system; A load characteristic similarity evaluation model and a fault prediction model are constructed based on real-time operation data and historical load data of the power distribution system. The load feature similarity assessment model is used to evaluate the similarity of load features between different regional groups, and the risk level of potential failure points is assessed by the failure prediction model. By performing correlation analysis on the load data of highly similar regions, high-risk fault points are marked and early warning information is generated.

[0008] Preferably, the cloud-based big data platform performs standardized data processing to establish real-time operation data tables and historical load data tables for the power distribution system, including the following steps: Establish data standardization rules to unify and transform various types of data collected by distributed data acquisition units into a standard format; The types of data collected by the distributed data acquisition unit include voltage data, current data, power data, and environmental data, among which environmental data includes temperature data and humidity data. The types of data collected by the distribution master station include the input and output power data of the main transformer, the grid frequency data, and the current fluctuation data of the main line. Establish a real-time operation data table including a data list of voltage data, current data, power data, and environmental data, and establish the association between data in the same area group based on the data identifiers in the data list; Establish a historical load data table that includes load peak data, load fluctuation data, and load distribution data for different time periods. Based on the timestamps in the data list, establish a comparative analysis link between historical data and real-time data.

[0009] Preferably, constructing a load feature similarity evaluation model includes the following steps: The load distribution characteristics are obtained based on the electrical topology data of the regional groups, and the voltage data, current data, power data and environmental data are collected by the distributed data acquisition unit; among which, the electrical topology data includes the connection relationship and number of nodes of the regional groups in the power distribution system topology diagram; Create an electrical topology data table and integrate it into a real-time operation data table; A comprehensive similarity recognition model for voltage and current data is constructed using deep learning algorithms. The comprehensive similarity of voltage and current data between region groups is evaluated using a comprehensive similarity identification model based on voltage and current data. Based on the comprehensive similarity assessment results, as well as the electrical topology data table, power data, and environmental data tables, a load characteristic similarity assessment model is established using formulas.

[0010] Preferably, constructing a fault prediction model includes the following steps: Acquire voltage, current, and environmental data collected by distributed data acquisition units in the power distribution system; The comprehensive similarity of voltage and current data between region groups is evaluated using a comprehensive similarity identification model based on voltage and current data. By extracting time-domain and frequency-domain features from environmental data of high-risk fault points in the power distribution system from historical data, and using machine learning algorithms to train an environmental data anomaly detection model to identify potential fault points; Potential fault points are identified by combining a similarity recognition model of voltage and current data with an anomaly detection model of environmental data.

[0011] Preferably, a comprehensive similarity recognition model for voltage and current data is constructed using a deep learning algorithm, including the following steps: The voltage and current data of highly similar regions in the power distribution system from historical data are organized into structured training and validation datasets. A convolutional neural network was trained using training and validation datasets to establish a comprehensive similarity recognition model for voltage and current data, identifying whether regions are highly similar.

[0012] Preferably, the load feature similarity evaluated by the load feature similarity evaluation model includes the following steps: The load characteristic similarity assessment value between region groups is calculated using a load characteristic similarity assessment model; If the load feature similarity evaluation value is greater than the preset threshold, the region group is a high similarity region group; otherwise, it is not a high similarity region group.

[0013] This invention also provides a smart power distribution management method based on big data, comprising the following steps: Distributed data acquisition units are set up at each key node of the power distribution system, and the nodes are divided into multiple area groups according to their physical location and electrical topology. Each area group is equipped with an area aggregation node to receive the data within that group. An edge computing chip and a wireless communication module are embedded in each distributed data acquisition unit. Each acquisition unit is uniquely numbered, and a data identifier is generated based on the timestamp of data acquisition and the node number. Low-latency communication links are established between adjacent acquisition units sequentially from the power distribution master station to the regional aggregation node. The distributed data acquisition units within the regional group send the collected real-time data to the regional aggregation node. The regional aggregation node performs preliminary cleaning and compression on the received data before packaging and transmitting it to the nearest backbone communication node. The backbone communication nodes dynamically monitor load data within and across regions, and adjust the load distribution strategy of each region based on historical load characteristics and real-time data through nonlinear optimization algorithms to achieve global load balancing. The power distribution master station receives data from the backbone communication nodes, performs comprehensive analysis of the data using a deep learning model to generate an operational status assessment report for the power distribution system, and provides precise early warnings for potential fault points.

[0014] Beneficial effects

[0015] This invention addresses the challenge of existing systems' inability to comprehensively analyze multi-source heterogeneous data. It achieves deep data mining from multiple sources through distributed acquisition and edge computing, combined with deep learning models for comprehensive analysis, thus fully reflecting the dynamic operating status of the power distribution system. It compensates for the lack of a dynamic load balancing mechanism by using nonlinear optimization algorithms to dynamically adjust load distribution, achieving global load balancing and improving system operating efficiency. It enhances the accuracy of fault early warning by constructing load feature similarity evaluation and fault prediction models, combining multi-dimensional data to identify potential fault points, thereby improving system reliability. Finally, it elevates the intelligent management level of the power distribution system, meeting the needs of modern smart grids for refined management and intelligent decision-making. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the structure of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Figure 1 The overall structure of a big data-based intelligent power distribution management system is demonstrated, including the connection relationships between the data acquisition module, communication network construction module, data preprocessing module, load balancing module, and intelligent decision-making module, as well as their distribution in the power distribution system.

[0019] like Figure 1As shown, the system first sets up distributed data acquisition units at each key node of the power distribution system, and divides these nodes into multiple regional groups based on their physical location and electrical topology. Each regional group has a regional aggregation node to receive data from that group. For example, assuming the power distribution system contains 100 key nodes, it can be divided into 10 regional groups based on the electrical connections between the nodes. Each regional group contains 10 nodes, and a regional aggregation node is assigned to each regional group, numbered R1 to R10.

[0020] An edge computing chip and a wireless communication module are embedded in each distributed data acquisition unit, and a unique device identifier is assigned to each unit. The main function of the edge computing chip is to perform preliminary processing of local data and anomaly detection, thereby reducing data transmission volume and improving system real-time performance. The choice of wireless communication module depends on the node's environment; for example, a 5G module is used in high-bandwidth scenarios, an NB-IoT module in low-power wide-area scenarios, and a Zigbee module in short-range communication scenarios. Furthermore, a power distribution system topology map is created based on the physical location and electrical connections of the nodes, recording the physical location, electrical connections, and the number of nodes along the shortest path to the power distribution master station for each acquisition unit. For example, acquisition unit C1 is located within area group R1, with a distance of 2 nodes from the area aggregation node R1 and a distance of 8 nodes from the power distribution master station.

[0021] Low-latency communication links are established sequentially between adjacent data acquisition units from the distribution master station to the regional aggregation node. This is achieved through wireless communication modules. Specifically, based on the distribution system topology, low-latency communication links between adjacent data acquisition units are established sequentially within the distribution network using wireless communication modules. For example, assuming data acquisition units C1 and C2 are adjacent within area group R1, a communication link between C1 and C2 is established using a wireless communication module. The regional aggregation node then establishes an efficient data transmission link with the nearest backbone communication node using the MQTT-SN protocol, ensuring the reliability and real-time performance of data transmission. For example, a link is established between regional aggregation node R1 and backbone communication node M1 using the MQTT-SN protocol to achieve efficient data transmission.

[0022] Distributed data acquisition units within a regional group send the collected real-time data to the regional aggregation node. The regional aggregation node performs preliminary cleaning and compression on the received data before packaging and transmitting it to the nearest backbone communication node. For example, assuming acquisition units C1 to C10 within regional group R1 collect voltage, current, power, and environmental data respectively, regional aggregation node R1 will perform preliminary cleaning on this data, removing outliers and noise, and reducing the data volume using a compression algorithm. The cleaned and compressed data is then packaged and transmitted to the backbone communication node M1. The backbone communication node dynamically monitors load data within and across regions and, based on historical load characteristics and real-time data, adjusts the load distribution strategy for each region using a nonlinear optimization algorithm to achieve global load balancing. The nonlinear optimization algorithm used by the backbone communication node aims to minimize the global load variance, with the objective function being: Where n is the total number of region groups. For the real-time load of the i-th region group, Let be the average load across all region groups. The objective function is solved using gradient descent, and the load distribution ratio for each region group is dynamically adjusted.

[0023] For example, assuming the load data of region groups R1 and R2 are L1 and L2 respectively, the backbone communication node will adjust the distribution ratio of L1 and L2 through a nonlinear optimization algorithm based on the historical load characteristics H1 and H2 and the real-time data R1 and R2, so that the global load reaches a balanced state.

[0024] The power distribution master station receives data from the backbone communication nodes, performs comprehensive analysis of the data through a deep learning model, generates an operational status assessment report for the power distribution system, and provides precise early warnings for potential fault points.

[0025] For example, the distribution master station aggregates the data transmitted by the backbone communication node M1 and uploads it to the cloud-based big data platform. The cloud-based big data platform standardizes the data, establishing real-time operation data tables and historical load data tables for the distribution system. Specifically, data standardization rules are formulated to uniformly convert various types of data collected by the distributed data acquisition units into standard formats. For example, voltage data, current data, power data, and environmental data are standardized into V1, I1, P1, and E1 formats, respectively, to facilitate subsequent analysis.

[0026] The distributed data acquisition units collect data including voltage, current, power, and environmental data, with environmental data including temperature and humidity. The distribution substation collects input and output power data of the main transformer, grid frequency data, and current fluctuation data of the main lines. A real-time operation data table is established, including lists of voltage, current, power, and environmental data. Based on the data identifiers in the data lists, relationships are established between data groups in the same area.

[0027] For example, data identifier D1 within region group R1 is associated with data identifier D2 within region group R2 to facilitate subsequent analysis. A historical load data table is created, including load peak data, load fluctuation data, and load distribution data for different time periods. Based on the timestamps in the data list, a comparative analysis link is established between historical and real-time data. For example, a comparative analysis link is established between timestamp T1 in the historical load data table and timestamp T2 in the real-time operation data table to identify load change trends.

[0028] Based on real-time operating data and historical load data of the power distribution system, a load characteristic similarity evaluation model and a fault prediction model are constructed.

[0029] For example, a load characteristic similarity assessment model can be used to evaluate the similarity of load characteristics between different regional groups, and a fault prediction model can be used to assess the risk level of potential fault points. Specifically, the process of constructing the load characteristic similarity assessment model is as follows: Load distribution characteristics are obtained based on the electrical topology data of the regional groups, including voltage, current, power, and environmental data collected by a distributed data acquisition unit. The electrical topology data includes the connection relationships and number of nodes of the regional groups in the power distribution system topology diagram. An electrical topology data table is established and integrated into a real-time operation data table. A comprehensive similarity recognition model for voltage and current data is constructed using a deep learning algorithm. The comprehensive similarity recognition model for voltage and current data is used to evaluate the comprehensive similarity of voltage and current data between regional groups. Based on the comprehensive similarity assessment results, the electrical topology data table, and the power and environmental data tables, the load characteristic similarity assessment model is established using the following formula: Electrical topology data similarity Calculation formula:

[0030] Power data similarity Calculation formula:

[0031] Environmental data similarity Calculation formula:

[0032] in, This is the load feature similarity evaluation value. To measure the similarity of electrical topology data between region groups. To measure the similarity of power data between regional groups To measure the similarity of environmental data between regional groups. A constant set based on the comprehensive similarity assessment results. , and The weights for electrical topology data, power data, and environmental data are respectively defined; and they satisfy the following conditions: The weights are determined by the entropy weight method, which assigns weights based on the information entropy value of each type of data. The smaller the entropy value, the richer the information the indicator carries, and the greater the weight, so as to improve the objectivity of the model evaluation. This represents the average number of nodes in the two region groups. This represents the difference in the number of nodes between the two region groups. This represents the average distance between nodes in the two region groups and the central node of the power distribution system topology. and These represent the distances between nodes in the two region groups and the central node, respectively. This represents the mean variance of power data in the two regional groups over a preset time period. This is the difference in variance of the power data. This represents the mean of the ambient temperature data in the two regional groups. This represents the difference in ambient temperature data.

[0033] For example, assuming the electrical topology data similarity between region groups R1 and R2 is 0.8, the power data similarity is 0.7, and the environmental data similarity is 0.6, with weights of 0.4, 0.3, and 0.3 respectively, and a constant C of 1.2, then the load characteristic similarity evaluation value S is: The process of constructing a fault prediction model is as follows: Voltage, current, and environmental data are acquired from distributed data acquisition units in the power distribution system. A comprehensive similarity identification model for voltage and current data is used to evaluate the overall similarity of voltage and current data between regional groups. Time-domain and frequency-domain features are extracted from environmental data of high-risk fault points in the power distribution system based on historical data. An environmental data anomaly detection model is then trained using machine learning algorithms to identify potential fault points.

[0034] For example, extracting time-domain features from environmental data includes the mean. ,variance and extreme values The time-domain signal is transformed using Fourier transform. Convert to frequency domain Extracting frequency domain features includes the main peak frequency. and total energy .

[0035] Potential fault points are identified by combining a similarity recognition model of voltage and current data with an anomaly detection model of environmental data: a comprehensive similarity threshold is set. (Recommended value: 0.8) and environmental anomaly risk threshold (A value of 0.7 is recommended). If the output value of the comprehensive similarity recognition model is ≥ And the risk level output by the environmental data anomaly detection model is ≥ If it is a potential fault point, it is considered a normal node; otherwise, it is considered a normal node.

[0036] For example, assuming the overall similarity between voltage and current data in region group R1 is 0.9, and the output of the environmental data anomaly detection model is high risk, then region group R1 is determined to be a potential fault point.

[0037] The process of constructing a comprehensive similarity recognition model for voltage and current data using deep learning algorithms is as follows: Voltage and current data from high-similarity regions in a power distribution system from historical data are organized into structured training and validation datasets. A convolutional neural network is trained using these datasets. This network contains three convolutional layers (using 32, 64, and 128 3×3 convolutional kernels respectively), two pooling layers (max pooling, 2×2 stride), one fully connected layer (128 neurons, ReLU activation function), and an output layer (one neuron, Sigmoid activation function, output value range [0,1], representing region group similarity). The training process uses the Adam optimizer with a learning rate of 0.001 and 50 iterations. Network parameters are adjusted using the validation set accuracy to establish a comprehensive similarity recognition model for voltage and current data, identifying whether regions are highly similar. For example, suppose the training dataset contains 1000 sets of voltage and current data, and the verification dataset contains 200 sets of voltage and current data. The model trained by the convolutional neural network can identify whether region groups R1 and R2 are highly similar region groups.

[0038] The process of evaluating load feature similarity using the load feature similarity evaluation model is as follows: The model calculates the load feature similarity evaluation value between region groups. If the load feature similarity evaluation value is greater than a preset threshold, the region group is considered a high-similarity region group; otherwise, it is not. For example, assuming the preset threshold is 0.8, and the load feature similarity evaluation value between region groups R1 and R2 is 0.852, then region groups R1 and R2 are considered high-similarity region groups.

[0039] A big data-based intelligent power distribution management method includes the following steps: Distributed data acquisition units are set up at key nodes of the power distribution system, and the nodes are divided into multiple regional groups according to their physical location and electrical topology. A regional aggregation node is set up within each regional group to receive data from that group. An edge computing chip and a wireless communication module are embedded in each distributed data acquisition unit. Each acquisition unit is uniquely numbered, and a data identifier is generated based on the data acquisition timestamp and node number. Low-latency communication links are established sequentially between adjacent acquisition units from the power distribution master station to the regional aggregation node. The distributed data acquisition units within a regional group send the collected real-time data to the regional aggregation node. The regional aggregation node performs preliminary cleaning and compression of the received data before packaging and transmitting it to the nearest backbone communication node. The backbone communication node dynamically monitors load data within and across regions, and adjusts the load distribution strategy for each region based on historical load characteristics and real-time data using a nonlinear optimization algorithm to achieve global load balancing. The power distribution master station receives data from the backbone communication node, performs comprehensive analysis of the data using a deep learning model, generates a power distribution system operation status assessment report, and provides precise early warnings for potential fault points.

[0040] As can be seen from the above embodiments, the intelligent power distribution management method and system based on big data provided by the present invention can effectively solve the problem that existing power distribution management systems are unable to perform in-depth mining and comprehensive analysis of multi-source heterogeneous data. At the same time, it has a dynamic load balancing compensation mechanism in complex scenarios, which significantly improves the operating efficiency and security of the power distribution system.

[0041] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0042] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart power distribution management system based on big data, characterized in that, include: Data acquisition module: Distributed data acquisition units are set up at each key node of the power distribution system, and the nodes are divided into multiple area groups according to their physical location and electrical topology. Each area group is equipped with an area aggregation node to receive the data within that group. Communication network construction module: An edge computing chip and a wireless communication module are embedded in each distributed data acquisition unit. The acquisition unit is uniquely numbered, and a data identifier is generated based on the data acquisition timestamp and node number. Low-latency communication links between adjacent acquisition units are established sequentially from the power distribution master station to the regional aggregation node. Data preprocessing module: The distributed data acquisition units within the regional group send the collected real-time data to the regional aggregation node. The regional aggregation node performs preliminary cleaning and compression on the received data, and then packages and transmits it to the nearest backbone communication node. Load balancing module: The backbone communication node dynamically monitors the load data within and across regions, and adjusts the load distribution strategy of each region based on historical load characteristics and real-time data through a nonlinear optimization algorithm to achieve global load balancing; Intelligent decision-making module: The power distribution master station receives data from the backbone communication nodes, performs comprehensive analysis of the data through a deep learning model to generate an operational status assessment report of the power distribution system, and provides accurate early warnings of potential fault points.

2. The system according to claim 1, characterized in that, An edge computing chip and a wireless communication module are embedded in each distributed data acquisition unit, and each acquisition unit is uniquely numbered, including the following steps: An edge computing chip and a wireless communication module are embedded in each distributed data acquisition unit. The edge computing chip is used to perform preliminary processing and anomaly detection of local data. A suitable data storage format is selected for efficient storage and subsequent analysis. The wireless communication module selects different communication technologies according to the environment in which the node is located. In high-bandwidth demand scenarios, a 5G module is used; in low-power wide-area scenarios, an NB-IoT module is used; and in short-range communication scenarios, a Zigbee module is used. Assign a unique device identifier to each distributed data acquisition unit and create a power distribution system topology map to record the physical location, electrical connections, and number of shortest path nodes between each acquisition unit and the power distribution master station.

3. The system according to claim 1, characterized in that, Establishing low-latency communication links between adjacent acquisition units sequentially from the power distribution master station to the regional aggregation node includes the following steps: From the main power distribution station to the regional aggregation node, low-latency communication links between adjacent acquisition units are established sequentially in the power distribution network through wireless communication modules according to the power distribution system topology. Regional aggregation nodes establish efficient data transmission links with the nearest backbone communication node via the MQTT-SN protocol to ensure the reliability and real-time performance of data transmission.

4. The system according to claim 1, characterized in that, The power distribution master station receives data from the backbone communication nodes, performs comprehensive analysis of the data using a deep learning model to generate an operational status assessment report for the power distribution system, and provides precise early warnings of potential fault points. This includes the following steps: The power distribution master station aggregates the data transmitted by the backbone communication nodes and uploads it to the cloud big data platform; the cloud big data platform performs standardized processing on the data to establish real-time operation data tables and historical load data tables for the power distribution system; A load characteristic similarity evaluation model and a fault prediction model are constructed based on real-time operation data and historical load data of the power distribution system. The load feature similarity assessment model is used to evaluate the similarity of load features between different regional groups, and the risk level of potential failure points is assessed by the failure prediction model. By performing correlation analysis on the load data of highly similar regions, high-risk fault points are marked and early warning information is generated.

5. The system according to claim 4, characterized in that, The cloud-based big data platform standardizes the data to establish real-time operation data tables and historical load data tables for the power distribution system, including the following steps: Establish data standardization rules to unify and transform various types of data collected by distributed data acquisition units into a standard format; The types of data collected by the distributed data acquisition unit include voltage data, current data, power data, and environmental data, among which environmental data includes temperature data and humidity data. The types of data collected by the distribution master station include the input and output power data of the main transformer, the grid frequency data, and the current fluctuation data of the main line. Establish a real-time operation data table including a data list of voltage data, current data, power data, and environmental data, and establish the association between data in the same area group based on the data identifiers in the data list; Establish a historical load data table that includes load peak data, load fluctuation data, and load distribution data for different time periods. Based on the timestamps in the data list, establish a comparative analysis link between historical data and real-time data.

6. The system according to claim 4, characterized in that, Constructing a load feature similarity evaluation model includes the following steps: The load distribution characteristics are obtained based on the electrical topology data of the regional groups, and the voltage data, current data, power data and environmental data are collected by the distributed data acquisition unit; among which, the electrical topology data includes the connection relationship and number of nodes of the regional groups in the power distribution system topology diagram; Create an electrical topology data table and integrate it into a real-time operation data table; A comprehensive similarity recognition model for voltage and current data is constructed using deep learning algorithms. The comprehensive similarity of voltage and current data between region groups is evaluated using a comprehensive similarity identification model based on voltage and current data. Based on the comprehensive similarity assessment results, as well as the electrical topology data table, power data, and environmental data tables, a load characteristic similarity assessment model is established using formulas.

7. The system according to claim 4, characterized in that, Constructing a fault prediction model includes the following steps: Acquire voltage, current, and environmental data collected by distributed data acquisition units in the power distribution system; The comprehensive similarity of voltage and current data between region groups is evaluated using a comprehensive similarity identification model based on voltage and current data. By extracting time-domain and frequency-domain features from environmental data of high-risk fault points in the power distribution system from historical data, and using machine learning algorithms to train an environmental data anomaly detection model to identify potential fault points; Potential fault points are identified by combining a similarity recognition model of voltage and current data with an anomaly detection model of environmental data.

8. The system according to claim 6, characterized in that, A comprehensive similarity recognition model for voltage and current data is constructed using deep learning algorithms, including the following steps: The voltage and current data of highly similar regions in the power distribution system from historical data are organized into structured training and validation datasets. A convolutional neural network was trained using training and validation datasets to establish a comprehensive similarity recognition model for voltage and current data, identifying whether regions are highly similar.

9. The system according to claim 6, characterized in that, The load feature similarity assessment using the load feature similarity evaluation model includes the following steps: The load characteristic similarity assessment value between region groups is calculated using a load characteristic similarity assessment model; If the load feature similarity evaluation value is greater than the preset threshold, the region group is a high similarity region group; otherwise, it is not a high similarity region group.

10. A smart power distribution management method based on big data, characterized in that, Includes the following steps: Distributed data acquisition units are set up at each key node of the power distribution system, and the nodes are divided into multiple area groups according to their physical location and electrical topology. Each area group is equipped with an area aggregation node to receive the data within that group. An edge computing chip and a wireless communication module are embedded in each distributed data acquisition unit. Each acquisition unit is uniquely numbered, and a data identifier is generated based on the timestamp of data acquisition and the node number. Low-latency communication links are established between adjacent acquisition units sequentially from the power distribution master station to the regional aggregation node. The distributed data acquisition units within the regional group send the collected real-time data to the regional aggregation node. The regional aggregation node performs preliminary cleaning and compression on the received data before packaging and transmitting it to the nearest backbone communication node. The backbone communication nodes dynamically monitor load data within and across regions, and adjust the load distribution strategy of each region based on historical load characteristics and real-time data through nonlinear optimization algorithms to achieve global load balancing. The power distribution master station receives data from the backbone communication nodes, performs comprehensive analysis of the data using a deep learning model to generate an operational status assessment report for the power distribution system, and provides precise early warnings for potential fault points.

Images