Intelligent power distribution network distributed power supply dispatching system based on edge calculation

Abstract

The invention relates to the technical field of power distribution network operation optimization, and discloses an intelligent power distribution network distributed power scheduling system based on edge computing, which comprises a data acquisition and preprocessing module, an edge computing and local scheduling module, an SDN network control module, an application layer global scheduling module and a scheduling execution and control module, according to the invention, by introducing the edge computing node, local preprocessing of data and generation of a preliminary scheduling strategy are realized, delay of data transmission to the central server is significantly reduced, and the system can dynamically adjust the scheduling priority in combination with the improved particle swarm optimization algorithm, so that the scheduling efficiency is improved. Three core indexes of renewable energy utilization rate, power distribution network loss and voltage stability are comprehensively considered, and a more accurate and efficient local scheduling strategy is generated. And meanwhile, the SDN network control module ensures that high-priority data can still be efficiently transmitted when the network is congested by calculating a link priority coefficient, so that the real-time performance and the reliability of the system are further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network operation optimization, and specifically to an intelligent distribution network distributed power supply scheduling system based on edge computing. Background Art

[0002] With the transformation of the global energy structure and the rapid development of renewable energy, the proportion of distributed power sources (such as solar energy, wind energy, etc.) in the distribution network is increasing. These distributed power sources have the characteristics of intermittency, volatility and uncertainty, which brings huge challenges to the operation and control of the distribution network. The scheduling method of the traditional distribution network mainly relies on centralized control strategies. All data needs to be transmitted to the central server for processing and decision-making, resulting in large data transmission delays and slow system response speeds, which makes it difficult to meet the real-time, reliability and flexibility requirements of modern smart distribution networks. In addition, with the expansion of the scale and increase of complexity of the distribution network, the computational burden of the centralized scheduling method has also increased sharply, further limiting the performance of the system. Therefore, the development of an efficient, real-time and reliable distributed power scheduling system has become a research hotspot in the current field of smart distribution networks.

[0003] Traditional distribution network dispatching technology has the following main deficiencies: First, the centralized dispatching mode leads to large data transmission delays and slow system response speed, which cannot meet real-time dispatching needs. In the context of large-scale access of distributed power sources, the real-time operating conditions of the distribution network change rapidly, and the centralized dispatching method is difficult to capture these changes in time and respond effectively. Secondly, the traditional dispatching method lacks comprehensive consideration of multiple objectives such as renewable energy utilization, distribution network loss and voltage stability, which often leads to conflicts between local optimization and global goals, affecting the overall performance of the system. Furthermore, the traditional network control method lacks intelligence in data transmission path selection and bandwidth allocation, making it difficult to ensure the transmission quality of high-priority data during network congestion. Finally, the traditional PID controller parameters are fixed and cannot be adaptively adjusted according to the real-time control deviation and deviation change rate, resulting in insufficient accuracy and stability in the operation control of distributed power sources. Summary of the Invention

[0004] The purpose of the present invention is to make up for the shortcomings of the existing technology and provide an intelligent distribution network distributed power supply scheduling system based on edge computing. The system realizes local data preprocessing and preliminary scheduling strategy generation by introducing edge computing nodes, significantly reducing data transmission delay. The system combines the improved particle swarm optimization algorithm, dynamically adjusts the scheduling priority, comprehensively considers the utilization rate of renewable energy, network loss and voltage stability, and the SDN network control module ensures the efficient transmission of high-priority data. The application layer global scheduling module resolves the conflict between local and global goals, realizes the seamless connection between global and local scheduling, and improves the real-time, reliability and flexibility of the system.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a distributed power supply dispatching system for an intelligent distribution network based on edge computing, the system comprising:

[0006] Data acquisition and preprocessing module: collects distributed power supply operating status data and distribution network real-time operating data through sensors, uses dynamic weighted outlier detection method to identify and eliminate outliers in the collected data, and transmits it to the edge computing node;

[0007] Edge computing and local scheduling module: The edge computing node pre-processes the received data and comprehensively optimizes renewable energy utilization, distribution network losses, and voltage stability based on the edge node collaborative scheduling objective function. It adaptively adjusts scheduling priorities in different scenarios using dynamic weight coefficients and uses an improved particle swarm optimization algorithm to generate a preliminary local scheduling strategy. The final local scheduling decision is made after information interaction and coordination with adjacent edge nodes.

[0008] SDN network control module: Receives local scheduling decisions, calculates link priority coefficients, quantitatively evaluates the priority of each transmission link, uses an improved Dijkstra algorithm to determine the optimal data transmission path, allocates network bandwidth based on a priority queue mechanism, monitors network traffic in real time, and handles congestion promptly when detected;

[0009] Application layer global scheduling module: Receives information from each edge node, evaluates the operating status of the distribution network, uses a mixed integer programming model to perform global optimization calculations, generates a global scheduling strategy, and sends it to the edge computing nodes;

[0010] Scheduling execution and control module: The edge computing node receives the global scheduling strategy, calculates the fusion coefficient based on the global-local scheduling fusion coefficient formula, generates the final execution instruction, controls the operation of the distributed power supply through the adaptive PID correction coefficient formula, and records the operation trajectory and uploads it to the application layer global scheduling module.

[0011] Furthermore, the sensors used in the data acquisition and preprocessing module are: sensors at the distributed power source, specifically the light sensor and photovoltaic power sensor on the photovoltaic panel, and the wind speed sensor, wind direction sensor, wind power monitoring sensor, and vibration sensor at the wind turbine; sensors on the power grid line, specifically current sensors, voltage sensors, and power quality analyzers; sensors on the energy storage system, specifically current sensors, voltage sensors, temperature sensors, and state of charge sensors.

[0012] Furthermore, the calculation formula for dynamic weighted outlier detection in the data acquisition and preprocessing module is: ,in, For the The outlier determination threshold for each sampling point is is the adaptive weight, For the past The standard deviation of the sampling points, is the current acquisition value of the kth sampling point, is a historical reference dataset, For the front The mean of the sampling points, is the sampling point window size, when The current acquisition value of the sampling point satisfy > When , the sampling value is determined to be an abnormal value. The abnormal value itself has no fixed value range, and its determination result changes dynamically with the fluctuation of real-time sampling data.

[0013] Furthermore, the calculation formula of the edge node collaborative scheduling objective function in the edge computing and local scheduling module is: ,in, is the scheduling objective function value, 、 、 is the weight coefficient, is the number of distributed power sources, For the The real-time output power of each distributed power source, For the The maximum output power of a distributed power supply, is the number of distribution network lines, For the Real-time power loss of the line, For the The maximum permissible loss of a line, is the number of distribution network nodes, For the The real-time voltage of each node, is the rated voltage of the node, is the maximum allowable voltage deviation.

[0014] Furthermore, the specific steps of improving the particle swarm optimization algorithm in the edge computing and local scheduling module are as follows: initializing the position and speed of the particle swarm, setting the number of iterations to 50, and the population size of each iteration to 30; in each iteration, calculating the fitness value of each particle, updating the individual optimal position of each particle and the global optimal position of the entire population; introducing an adaptive inertia weight, with the initial inertia weight set to 0.9, which linearly decreases to 0.4 as the number of iterations increases, and adjusting the speed and position of the particles according to the updated inertia weight; repeating the iterative process until the number of iterations reaches 50, at which point the solution corresponding to the global optimal position is the local preliminary scheduling strategy.

[0015] Furthermore, the calculation formula for calculating the link priority coefficient when determining the data transmission path in the SDN network control module is: ,in, is the dynamic priority coefficient of the link, 、 、 is the weight coefficient, is the data value coefficient, is the current load of the link, is the maximum load of the link, is the link reliability index.

[0016] Furthermore, in the SDN network control module, the priority queue mechanism is a mechanism that divides different priority queues according to the importance and real-time requirements of data transmission and allocates network resources according to priority. Data transmission is divided into three levels, with scheduling instructions as the highest priority, real-time monitoring data as medium priority, and historical statistical data as low priority, using the remaining bandwidth.

[0017] Furthermore, the calculation formula of the global-local scheduling fusion coefficient in the scheduling execution and control module is: ,in is the global-local scheduling fusion coefficient, is the confidence of the global scheduling model, is the global optimization gain coefficient, is the confidence of the local edge node, is the local response speed coefficient.

[0018] Furthermore, the specific steps of generating the final instruction from the global-local scheduling fusion coefficient in the scheduling execution and control module are as follows:

[0019] (1) The edge computing node receives the global scheduling policy issued by the application layer and the locally generated scheduling decision;

[0020] (2) Calculate the global-local scheduling fusion coefficient ;

[0021] (3) Using the fusion coefficient β as the weight, the instruction value in the global scheduling strategy and the instruction value in the local scheduling decision are weighted and fused. That is, the weight of the global instruction value is β, and the weight of the local instruction value is 1-β. The two are superimposed and calculated according to this weight ratio to obtain the final execution instruction;

[0022] (4) Perform format check on the generated final execution instruction to ensure that it complies with the instruction format requirements of the distributed power supply controller;

[0023] (5) The final execution instruction is transmitted to the corresponding distributed power supply controller through the industrial bus to control the operating status of the distributed power supply.

[0024] Furthermore, the calculation formula of the adaptive PID correction coefficient in the scheduling execution and control module is: , ,in 、 、 They are The proportional coefficient, integral coefficient, and differential coefficient at each moment, 、 、 are the initial proportional coefficient, integral coefficient, and differential coefficient respectively, for The control deviation at the moment, is the rate of change of deviation at time t, is the maximum allowable control deviation, is the maximum allowable deviation change rate, 、 In order to adjust the coefficient, the actual operating status of the distributed power supply is tracked to the final execution instruction by adjusting the PID controller parameters in real time.

[0025] Compared with existing technologies, this intelligent distribution network distributed power dispatching system based on edge computing has the following beneficial effects:

[0026] 1. By introducing edge computing nodes, the present invention achieves local data preprocessing and preliminary scheduling strategy generation, significantly reducing the delay in data transmission to the central server. Combined with the improved particle swarm optimization algorithm, the system can dynamically adjust the scheduling priority, comprehensively considering the three core indicators of renewable energy utilization, distribution network loss, and voltage stability, to generate a more accurate and efficient local scheduling strategy. At the same time, the SDN network control module calculates the link priority coefficient to ensure that high-priority data can still be efficiently transmitted during network congestion, further improving the real-time performance and reliability of the system and making distributed power scheduling more flexible and rapid.

[0027] 2. The present invention receives information from each edge node through the application layer global scheduling module and adopts a mixed integer programming model to perform global optimization calculations, effectively solving the problem of conflict between local optimization and global goals in traditional methods. In addition, the scheduling execution and control module organically combines the global scheduling strategy with the local scheduling decision through the global-local scheduling fusion coefficient formula, generates the final execution instruction, and realizes the seamless connection between global and local scheduling. At the same time, the introduction of the adaptive PID correction coefficient formula enables the system to automatically adjust the PID controller parameters according to the real-time control deviation and the deviation change rate, thereby improving the accuracy and stability of the distributed power supply operation control, and providing a strong guarantee for the safe and efficient operation of the smart distribution network.

[0028] Other advantages, objects and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be learned from the practice of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0030] Figure 1 This is the workflow diagram of the distributed power dispatching system for smart distribution networks based on edge computing.

[0031] Figure 2 This is a diagram of the module architecture and data interaction of the distributed power dispatching system for smart distribution networks based on edge computing. DETAILED DESCRIPTION

[0032] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0033] Example 1: Smart distribution network scheduling scenario in an urban industrial park

[0034] Data acquisition and preprocessing module: Staff installed a large number of high-precision sensors at key locations of photovoltaic arrays, gas turbine inlet and outlet pipelines, distribution network transmission lines, and connection points of various electrical equipment. The sensors at distributed power sources include light sensors and photovoltaic power sensors on photovoltaic panels, as well as wind speed sensors, wind direction sensors, wind power monitoring sensors, and vibration sensors at wind turbines; sensors on power grid lines include current sensors, voltage sensors, and power quality analyzers; sensors on energy storage systems include current sensors, voltage sensors, temperature sensors, and state of charge sensors. These sensors can capture data such as the actual output power of photovoltaic power stations, the operating temperature and speed of gas turbines, the real-time voltage and current values ​​of distribution networks, the power loss of each line, and the real-time load of electrical equipment in real time. The data acquisition and preprocessing module is then started, using the dynamic weight outlier detection method to screen these massive data. The method is based on the calculation formula of outliers: ,in, For the The outlier determination threshold for each sampling point is is the adaptive weight, For the past The standard deviation of the sampling points, is the current acquisition value of the kth sampling point, is a historical reference dataset, For the front The mean of the sampling points, The sampling point window size is used to identify abnormal values ​​by calculating the degree to which the data deviates from the normal range. For example, when a sensor sends a value far beyond the normal range due to a sudden failure, this method can quickly identify this as abnormal data and remove it, ensuring that the data transmitted to the edge computing node is true and valid, laying a solid foundation for subsequent scheduling work. Figure 1 shown.

[0035] Edge computing and local scheduling module: The industrial park has divided multiple edge computing nodes into different functional areas, such as production areas, office areas, and storage areas. Each edge computing node acts as a "regional commander." After receiving pre-processed data, it first filters the data to remove interference noise and then performs standardization to enable comparison and analysis of data of different types and units under the same standards. The objective function of collaborative scheduling based on the edge nodes is calculated as follows: ,in, is the scheduling objective function value, 、 、 is the weight coefficient, is the number of distributed power sources, For the The real-time output power of each distributed power source, For the The maximum output power of a distributed power supply, is the number of distribution network lines, For the Real-time power loss of the line, For the The maximum permissible loss of a line, is the number of distribution network nodes, For the The real-time voltage of each node, is the rated voltage of the node, The function integrates multiple evaluation metrics, including the utilization of renewable energy sources such as photovoltaics (maximizing the effective use of photovoltaic power to minimize waste), the loss of each line in the distribution network (minimizing power loss during transmission), and the voltage stability at each node (ensuring the normal operation of electrical equipment). Dynamic weight coefficient formulas adjust the priority of these three metrics. For example, during peak industrial production hours, voltage stability is prioritized to prevent voltage fluctuations from impacting production. Next, an improved particle swarm optimization algorithm, combined with current power load and energy output, generates a preliminary local scheduling strategy. For example, this strategy adjusts the proportion of power supplied to the grid by a photovoltaic power station in a certain area or determines whether a gas turbine should be started or shut down. Subsequently, each edge computing node exchanges information with nodes in adjacent areas. For example, nodes in a production area communicate with nodes in an adjacent storage area about their respective scheduling plans, coordinating energy distribution and equipment operating status across different areas. Ultimately, a comprehensive set of local scheduling decisions is formed and transmitted to the SDN network control module.

[0036] SDN network control module: After the SDN control layer receives the data transmission request from each edge computing node, it immediately starts to calculate the link priority coefficient. The calculation formula of this coefficient is: ,in, is the dynamic priority coefficient of the link, 、 、 is the weight coefficient, is the data value coefficient, is the current load of the link, is the maximum load of the link, The link reliability index is a process that is like "scoring" different transmission links. Scoring is based on the importance of the data (for example, data such as scheduling instructions that can directly affect the operation of the equipment are much more valuable than ordinary monitoring data), the current link load (if a link has transmitted a large amount of data and is close to full load, its priority will be reduced), and the link's transmission reliability (some links may be prone to errors when transmitting data due to reasons such as line aging, and their priority will also be affected). Based on these "scores", the priority of each transmission link is evaluated to provide a basis for subsequent path selection and bandwidth allocation. Then, the improved The Dijkstra algorithm is used to select the optimal path from many possible transmission paths. This path not only has a fast transmission speed but also high reliability. At the same time, network bandwidth is allocated based on the priority queue mechanism, so that high-priority data (such as scheduling instructions) can obtain more bandwidth resources and be transmitted first. During the entire data transmission process, the SDN control layer also monitors network traffic in real time. Once it is found that a certain line is congested due to excessive data transmission, it will quickly take measures, such as switching part of the data to other less loaded paths, or adjusting the bandwidth allocation ratio to ensure that all data can be transmitted safely and efficiently to the destination.

[0037] Application layer global scheduling module: The application layer is like the "headquarters" of the entire industrial park's distribution network. It collects local scheduling decisions from all edge computing nodes, as well as the overall operating data of the distribution network, such as the total power load of the entire park, the total output of renewable energy, and the loss distribution of each line. Through comprehensive analysis of this information, it evaluates the current overall operating status of the distribution network. Afterwards, a mixed integer programming model is used for global optimization calculations. This model takes the global optimization objective function as its core. From the perspective of the entire industrial park, it balances the power output and power load of each area to avoid overload of local lines due to excessive burden. For example, when it is found that the line loss in a certain area is too high, a global scheduling strategy will be generated to adjust the output of the gas turbines around the area to supplement the power demand in the area and reduce the transmission pressure of the line. These global scheduling strategies are then distributed to each edge computing node.

[0038] Scheduling execution and control module: After the edge computing node receives the global scheduling policy issued by the application layer, it calculates the global-local scheduling fusion coefficient , the coefficient is calculated as: ,in is the global-local scheduling fusion coefficient, is the confidence of the global scheduling model, is the global optimization gain coefficient, is the confidence of the local edge node, is the local response speed coefficient. This coefficient is like a "regulator" to balance the weights of the global scheduling strategy and the local scheduling decision. According to this fusion coefficient, the global strategy and local decision are organically combined to generate the final execution instruction. In order to make the actual operating status of the distributed power supply accurately operate according to the final execution instruction, the system calculates the adaptive PID correction coefficient. The calculation formula is: , ,in 、 、 They are The proportional coefficient, integral coefficient, and differential coefficient at each moment, 、 、 are the initial proportional coefficient, integral coefficient, and differential coefficient respectively, for The control deviation at the moment, is the rate of change of deviation at time t, is the maximum allowable control deviation, is the maximum allowable deviation change rate, 、 In order to adjust the coefficient, the parameters of the PID controller in equipment such as photovoltaic inverters and gas turbine controllers are dynamically adjusted. For example, when the actual output power of the photovoltaic system deviates from the instruction requirements, the inverter parameters will be adjusted in time to ensure that the photovoltaic output meets the instruction requirements. At the same time, the system will record the operation trajectory of each adjustment in detail, such as how much the output of the photovoltaic power station was adjusted within a certain period of time, when the gas turbine was started or stopped, etc., and upload these records to the application layer to facilitate subsequent inquiries and analysis by staff.

[0039] To sum up, in the smart distribution network scheduling scenario of urban industrial parks, the system of the present invention effectively responds to the complex power consumption environment and energy supply situation in the park through the collaborative work of various modules. The data acquisition and preprocessing module ensures the accuracy of basic data and provides a reliable basis for subsequent scheduling; the edge computing and local scheduling module flexibly adjusts the scheduling priority based on actual needs, generates local strategies that fit the actual situation of the region, and realizes collaborative scheduling between regions through coordination with adjacent nodes; the SDN network control module ensures the efficient and secure transmission of various types of data, especially the priority transmission of key scheduling instructions; the application layer global scheduling module optimizes resource allocation from the perspective of the entire park and avoids problems that may be caused by local scheduling; the scheduling execution and control module ensures that distributed power sources operate accurately according to instructions by integrating global and local strategies and combining adaptive PID correction. The entire system significantly improves the real-time, reliability and flexibility of the industrial park distribution network, realizes the efficient use of renewable energy, reduces distribution network losses and stabilizes voltage, and provides strong support for stable production and efficient energy utilization in industrial parks.

[0040] Example 2: Rural Distributed Microgrid Dispatching Scenario

[0041] Data collection and preprocessing: Sensors are installed near the blades of wind turbines in wind farms, at the water inlets and outlets of hydropower station turbines, on the connection lines of energy storage battery packs, and on various branches of the distribution network. The sensors at the distributed power supply are specifically the light sensors and photovoltaic power sensors on the photovoltaic panels, and the wind speed sensors, wind direction sensors, wind power monitoring sensors, and vibration sensors at the wind turbines; the sensors on the power grid are specifically current sensors, voltage sensors, and power quality analyzers; the sensors on the energy storage system are specifically current sensors, voltage sensors, temperature sensors, and state of charge sensors. These sensors continuously collect data such as the actual output of wind power, the speed of water flow in the hydropower station, the state of charge of the energy storage battery (that is, the proportion of the current battery power to the total capacity), the voltage fluctuations of various branches of the distribution network, and the real-time electricity load of residents. After the data collection and preprocessing module is started, the dynamic weighted outlier detection method is used to process these data. The calculation formula based on the outlier is: , can effectively identify data that deviates from the normal range. For example, in heavy rain, some sensors may malfunction due to water ingress and send out some unreasonable data. This method can promptly identify these abnormal data and remove them, ensuring that the data transmitted to the edge computing node is accurate and available. Figure 2 shown.

[0042] Edge computing and local scheduling: Based on the distribution of villages in the countryside, the entire microgrid is divided into multiple edge computing nodes. Each node is responsible for energy scheduling for one or several adjacent villages. Each edge computing node receives pre-processed data and first performs filtering to remove noise caused by slight vibrations of the equipment. Then, it performs standardization to unify the format and range of the data. Based on the edge node collaborative scheduling objective function, the objective function is calculated as follows: This function integrates a variety of key factors, taking into account the utilization efficiency of renewable energy sources such as wind power and hydropower (try to make wind energy and hydropower fully utilized and not wasted), the loss of distribution network lines (reduce the loss of electricity in the process of transmitting from power generation equipment to residents' homes) and the voltage stability of each node (ensure that the electrical appliances in residents' homes can be used normally). The priority of these three indicators is adjusted through the dynamic weight coefficient formula. For example, in the dry season, the power generation of hydropower stations will decrease. At this time, the utilization rate of hydropower will be increased first, so that every drop of water can play the greatest role. Then, the improved particle swarm optimization algorithm is used, combined with the current renewable energy output, the power of energy storage batteries and The system generates preliminary local scheduling strategies based on residents' electricity consumption. For example, it determines the time period for charging the energy storage battery (when wind power or hydropower output is high) and the time period for discharging the energy storage battery (when wind power or hydropower output is insufficient), and adjusts the proportion of wind power connected to the distribution network (rationally allocates the wind power transmission capacity according to the size of the electricity load). After that, each edge computing node will exchange information with the nodes in the adjacent villages. For example, the nodes in one village will share their respective energy storage battery status and electricity demand with the nodes in the neighboring village, coordinate the power distribution of cross-village lines, avoid problems in a certain line due to excessive load, and finally form the final local scheduling decision and send it to the SDN network control module.

[0043] SDN network control: After receiving the data transmission request from each edge computing node, the SDN control layer starts to calculate the link priority coefficient. The calculation formula of this coefficient is: During the calculation process, the value of the data (data that directly affects energy storage and supply, such as the charging and discharging instructions of energy storage batteries, are of relatively high value), the load of the current link (for those lines connecting remote villages with weak transmission capabilities, they will be prevented from taking on excessive transmission tasks), and the reliability of the link (in harsh environments such as thunderstorms, priority should be given to lines with strong anti-interference capabilities and more stable transmission). Factors such as these are integrated to evaluate the priority of each transmission link. Then, the improved Dijkstra algorithm is used to select the most suitable path from many possible transmission paths. This path must be able to ensure the speed and stability of data transmission. Network bandwidth is allocated based on the priority queue mechanism, so that important data can be transmitted first and will not be delayed due to network congestion. At the same time, the network's operating status is monitored in real time. Once it is discovered that a certain communication line has unstable signals due to strong winds or other reasons, which may affect data transmission, it will quickly switch to other more stable transmission paths to ensure that data can be transmitted continuously and securely.

[0044] Application layer global scheduling: The application layer collects local scheduling decisions uploaded by all edge computing nodes and the operating data of the entire rural microgrid, including the total output of wind power and hydropower, the total capacity and current total power of energy storage batteries, the total power load of residents, and other information, comprehensively evaluates the overall operating status of the microgrid, and then calculates the global-local scheduling fusion coefficient. , the coefficient is calculated as: The model revolves around the global optimization objective function and coordinates the cooperation between wind power, hydropower and energy storage batteries from the perspective of the entire village. For example, when the wind is strong, the output of wind power will increase significantly. At this time, the energy storage batteries will be charged first to store the excess electricity; when the wind is weak, the energy storage batteries will be discharged to supplement the power supply. Based on these optimization results, a global scheduling strategy is generated and sent to each edge computing node.

[0045] Scheduling execution and control: After receiving the global scheduling policy issued by the application layer, the edge computing node calculates the fusion coefficient based on the global-local scheduling fusion coefficient formula, fuses the global scheduling policy with the locally generated scheduling decision, and forms the final execution instruction. In order to ensure that the actual operating status of the distributed power supply can accurately follow the final execution instruction, the system uses the adaptive PID correction coefficient formula, which is calculated as follows: , , dynamically adjust the parameters of the wind turbine's pitch angle controller (changing the wind power output by adjusting the angle of the blades), the hydropower station's turbine speed governor (controlling the turbine's speed to adjust the hydropower output), and the energy storage battery's charge and discharge controller. For example, when the final execution instruction requires an increase in wind power output, the parameters of the pitch angle controller will be adjusted to allow the blades to better capture wind energy and increase power generation. At the same time, the system will record the trajectory of each operation in detail, such as the charge and discharge amount of the energy storage battery in a certain time period, the adjustment time and amplitude of the wind turbine's pitch angle, etc., and upload these records to the application layer to facilitate staff to analyze and summarize the operation of the microgrid.

[0046] To sum up, in the rural distributed microgrid scheduling scenario, the system of the present invention fully adapts to the characteristics of renewable energy in rural areas that are greatly affected by natural conditions and electricity consumption is dispersed. The data acquisition and preprocessing module effectively eliminates abnormal data and ensures data quality; the edge computing and local scheduling module performs regional management according to the distribution of villages, adjusts the scheduling priority based on the energy characteristics of different time periods, and achieves energy balance between villages through coordination of adjacent nodes; the SDN network control module optimizes the data transmission path and bandwidth allocation based on the characteristics of rural communication lines, ensuring stable data transmission in severe weather; the application layer global scheduling module coordinates the cooperation of wind power, hydropower and energy storage from the overall perspective of the village, thereby improving the overall energy utilization efficiency; the scheduling execution and control module integrates global and local strategies and uses adaptive PID correction to enable distributed power sources to accurately respond to scheduling instructions. The system effectively solves the problems of insufficient utilization of renewable energy and unstable voltage in rural microgrids, improves the operating efficiency and reliability of microgrids, and provides a solid guarantee for the stable electricity use of rural residents and the promotion and application of clean energy.

[0047] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as above in terms of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, make some changes or modifications to equivalent embodiments using the technical contents disclosed above. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. The distributed power dispatching system of smart distribution network based on edge computing is characterized by: The system includes: Data acquisition and preprocessing module: collects distributed power supply operating status data and distribution network real-time operating data through sensors, uses dynamic weighted outlier detection method to identify and eliminate outliers in the collected data, and transmits it to the edge computing node; Edge computing and local scheduling module: The edge computing node pre-processes the received data and comprehensively optimizes renewable energy utilization, distribution network losses, and voltage stability based on the edge node collaborative scheduling objective function. It adaptively adjusts scheduling priorities in different scenarios using dynamic weight coefficients and uses an improved particle swarm optimization algorithm to generate a preliminary local scheduling strategy. The final local scheduling decision is made after information interaction and coordination with adjacent edge nodes. SDN network control module: Receives local scheduling decisions, calculates link priority coefficients, quantitatively evaluates the priority of each transmission link, uses an improved Dijkstra algorithm to determine the optimal data transmission path, allocates network bandwidth based on a priority queue mechanism, monitors network traffic in real time, and handles congestion promptly when detected; Application layer global scheduling module: Receives information from each edge node, evaluates the operating status of the distribution network, uses a mixed integer programming model to perform global optimization calculations, generates a global scheduling strategy, and sends it to the edge computing nodes; Scheduling execution and control module: The edge computing node receives the global scheduling strategy, calculates the fusion coefficient based on the global-local scheduling fusion coefficient formula, generates the final execution instruction, controls the operation of the distributed power supply through the adaptive PID correction coefficient formula, and records the operation trajectory and uploads it to the application layer global scheduling module.

2. The distributed power dispatching system for smart distribution network based on edge computing according to claim 1 is characterized in that: The sensors used in the data acquisition and preprocessing module are: sensors at the distributed power source, specifically the light sensor and photovoltaic power sensor on the photovoltaic panel, and the wind speed sensor, wind direction sensor, wind power monitoring sensor, and vibration sensor at the wind turbine; sensors on the power grid line, specifically current sensors, voltage sensors, and power quality analyzers; sensors on the energy storage system, specifically current sensors, voltage sensors, temperature sensors, and state of charge sensors.

3. The distributed power dispatching system for smart distribution network based on edge computing according to claim 1, characterized in that: The calculation formula for dynamic weight outlier detection in the data acquisition and preprocessing module is: ,in, For the The outlier determination threshold for each sampling point is is the adaptive weight, For the past The standard deviation of the sampling points, is the current acquisition value of the k-th sampling point, is a historical reference dataset, For the front The mean of the sampling points, is the sampling point window size, when The current acquisition value of the sampling point satisfy > When , the sampling value is determined to be an abnormal value. The abnormal value itself has no fixed value range, and its determination result changes dynamically with the fluctuation of real-time sampling data.

4. The distributed power dispatching system for smart distribution network based on edge computing according to claim 1, characterized in that: The calculation formula of the edge node collaborative scheduling objective function in the edge computing and local scheduling module is: ,in, is the scheduling objective function value, 、 、 is the weight coefficient, is the number of distributed power sources, For the The real-time output power of each distributed power source, For the The maximum output power of a distributed power supply, is the number of distribution network lines, For the Real-time power loss of the line, For the The maximum permissible loss of a line, is the number of distribution network nodes, For the The real-time voltage of each node, is the rated voltage of the node, is the maximum allowable voltage deviation.

5. The distributed power dispatching system for smart distribution network based on edge computing according to claim 1, characterized in that: The specific steps of improving the particle swarm optimization algorithm in the edge computing and local scheduling module are as follows: initializing the position and speed of the particle swarm, setting the number of iterations to 50, and the population size of each iteration to 30; in each iteration, calculating the fitness value of each particle, updating the individual optimal position of each particle and the global optimal position of the entire population; introducing an adaptive inertia weight, with the initial inertia weight set to 0.9 and linearly decreasing to 0.4 as the number of iterations increases, and adjusting the speed and position of the particles according to the updated inertia weight; repeating the iterative process until the number of iterations reaches 50, at which point the solution corresponding to the global optimal position is the local preliminary scheduling strategy.

6. The distributed power dispatching system for smart distribution network based on edge computing according to claim 1, characterized in that: The calculation formula for calculating the link priority coefficient when determining the data transmission path in the SDN network control module is: ,in, is the dynamic priority coefficient of the link, 、 、 is the weight coefficient, is the data value coefficient, is the current load of the link, is the maximum load of the link, is the link reliability index.

7. The distributed power dispatching system for smart distribution network based on edge computing according to claim 1, characterized in that: In the SDN network control module, the priority queue mechanism is a mechanism that divides different priority queues according to the importance and real-time requirements of data transmission and allocates network resources according to priority. Data transmission is divided into three levels, with scheduling instructions as the highest priority, real-time monitoring data as the medium priority, and historical statistical data as the low priority, using the remaining bandwidth.

8. The distributed power dispatching system for smart distribution network based on edge computing according to claim 1, characterized in that: The calculation formula of the global-local scheduling fusion coefficient in the scheduling execution and control module is: ,in is the global-local scheduling fusion coefficient, is the confidence of the global scheduling model, is the global optimization gain coefficient, is the confidence of the local edge node, is the local response speed coefficient.

9. The distributed power dispatching system for smart distribution network based on edge computing according to claim 1, characterized in that: The specific steps of generating the final instruction by the global-local scheduling fusion coefficient in the scheduling execution and control module are as follows: (1) The edge computing node receives the global scheduling policy issued by the application layer and the locally generated scheduling decision; (2) Calculate the global-local scheduling fusion coefficient ; (3) Using the fusion coefficient β as the weight, the instruction value in the global scheduling strategy and the instruction value in the local scheduling decision are weighted and fused. That is, the weight of the global instruction value is β, and the weight of the local instruction value is 1-β. The two are superimposed and calculated according to this weight ratio to obtain the final execution instruction; (4) Perform format check on the generated final execution instruction to ensure that it complies with the instruction format requirements of the distributed power supply controller; (5) The final execution instruction is transmitted to the corresponding distributed power supply controller through the industrial bus to control the operating status of the distributed power supply.

10. The distributed power dispatching system for smart distribution network based on edge computing according to claim 1, characterized in that: The calculation formula of the adaptive PID correction coefficient in the scheduling execution and control module is: , ,in 、 、 They are The proportional coefficient, integral coefficient, and differential coefficient at each moment, 、 、 are the initial proportional coefficient, integral coefficient, and differential coefficient respectively, for The control deviation at the moment, is the rate of change of deviation at time t, is the maximum allowable control deviation, is the maximum allowable deviation change rate, 、 In order to adjust the coefficient, the actual operating status of the distributed power supply is tracked to the final execution instruction by adjusting the PID controller parameters in real time.

Images