Distributed collaborative optimization control method, system, equipment and medium for distribution network

Abstract

The present invention relates to the field of intelligent optimization operation of distribution networks, and discloses a distributed collaborative optimization control method, system, equipment and medium for distribution networks. The method divides the distribution network into zones and configures edge intelligent agents to monitor the network loss data of the distribution network in real time. When the network loss exceeds a preset threshold, an optimization control process is triggered: with the network active loss as the objective function, the line flow, distributed power supply power and node voltage are constrained and reconstructed through semi-definite programming to construct an optimization control model as a convex quadratic programming model; based on the coupling branch states between each area, the constructed model is decomposed into several distributed optimization control sub-models based on zone coordination, and a synchronous alternating direction multiplier algorithm that introduces historical iteration results and increments of each edge intelligent agent is used to solve each sub-model, obtain a distributed collaborative optimization strategy and execute it; through zone management, distributed optimization and collaborative control, efficient operation and optimization of the distribution network are achieved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent optimization operation of distribution networks, and in particular to a distributed collaborative optimization control method, system, equipment and medium for distribution networks. Background Art

[0002] As the terminal link of power transmission, the operating stability and loss of the distribution network are directly related to the economy and reliability of the entire power grid.

[0003] Traditional distribution network operation optimization methods often rely on centralized control systems, which use central processing units to collect, analyze, and process global data to formulate optimization strategies. However, as distribution networks continue to expand in size and complexity, centralized control methods have gradually exposed problems such as large data communication volumes, heavy computational burdens, and slow response speeds. These methods are unable to meet the high real-time and reliability requirements of modern distribution networks, resulting in increasingly poor operational stability.

[0004] It can be seen that how to effectively improve the operational stability of the distribution network has become a technical problem that needs to be urgently solved by those skilled in the art. Summary of the Invention

[0005] The present invention provides a distributed collaborative optimization control method, system, device and medium for a distribution network, which solves the problem of how to effectively improve the operational stability of the distribution network.

[0006] To solve the above technical problems, the present invention provides a distributed collaborative optimization control method for a distribution network, comprising:

[0007] Real-time monitoring of distribution network loss data;

[0008] When the network loss data exceeds a preset network loss threshold, the network active power loss of the distribution network is used as an objective function, the line flow, distributed power source power and node voltage in the distribution network are constrained respectively, and the resulting constraint conditions are reconstructed by semi-definite programming. The optimization control model of the distribution network is constructed based on the reconstruction results and the objective function;

[0009] Acquiring coupling branch states between a plurality of edge areas in the distribution network, and decomposing the optimization control model into a plurality of distributed optimization control sub-models based on partition coordination based on the coupling branch states;

[0010] The edge intelligent agents deployed in each of the edge areas are controlled to use a preset distributed optimization algorithm to solve each of the distributed optimization control sub-models to obtain a distributed collaborative optimization strategy; the preset distributed optimization algorithm is a synchronous alternating direction multiplier algorithm that introduces the historical iteration results and increments of each of the edge intelligent agents.

[0011] Compared with the prior art, the embodiments of the present invention have the following advantages:

[0012] (1) By monitoring network loss data in real time through edge intelligent agents, network loss anomalies can be quickly discovered and optimized control can be triggered, thereby improving the stability and reliability of the distribution network. Through semi-definite programming reconstruction, the optimization control model is transformed into a convex quadratic programming model, ensuring the global optimality and computational efficiency of the solution, thereby improving the accuracy and efficiency of network loss calculation.

[0013] (2) The global optimization problem is decomposed into multiple sub-problems and solved by a distributed optimization algorithm, which reduces the computational complexity and improves the computational efficiency. The partition coordination mechanism based on the coupled branch state can effectively handle the interaction between edge areas and ensure global consistency. The historical iteration results and incremental information are introduced in the solution process to accelerate the convergence of the algorithm and improve the optimization performance. The present invention realizes the efficient operation and optimization of the distribution network through partition management, distributed optimization and collaborative control. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for use in the implementation. Obviously, the drawings described below are only some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0015] Figure 1 This is a flow chart of a distributed collaborative optimization control method for a distribution network provided by an embodiment of the present invention;

[0016] Figure 2 This is a schematic diagram of the division of the edge area of ​​the distribution network provided by an embodiment of the present invention;

[0017] Figure 3 is a flowchart for solving each distributed optimization control sub-model provided by an embodiment of the present invention;

[0018] Figure 4 A schematic diagram of the IEEE 33-node system topology and edge partitioning provided in an embodiment of the present invention;

[0019] Figure 5 A voltage comparison diagram of an IEEE 33-bus system before and after optimization provided by an embodiment of the present invention;

[0020] Figure 6 A boundary residual iterative change graph of each edge of an IEEE 33-node system provided by an embodiment of the present invention;

[0021] Figure 7A reactive output iterative change diagram of a distributed power source in an IEEE 33-node system provided by an embodiment of the present invention;

[0022] Figure 8 A daily load curve diagram of an IEEE 33-bus system provided in accordance with an embodiment of the present invention;

[0023] Figure 9 A comparison chart of network losses before and after optimization of an IEEE 33-node system according to an embodiment of the present invention;

[0024] Figure 10 This is a structural diagram of a distributed collaborative optimization control system for a distribution network provided by an embodiment of the present invention;

[0025] Figure 11 This is a structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The following is a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings and embodiments. Obviously, the embodiments described are only some embodiments of the present invention, rather than all embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0027] In the description of this application, the terms "first," "second," "third," etc. are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first," "second," "third," etc. may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0028] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be a communication between the two components. The terms "vertical", "horizontal", "left", "right", "up", "down" and similar expressions used herein are for illustrative purposes only, and do not indicate or imply that the system or component referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to the specific circumstances.

[0029] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this application have the same meanings as those commonly understood by those skilled in the art. The terms used in this specification are only for describing specific embodiments and are not intended to limit the present invention. Those skilled in the art will understand the specific meanings of the above terms in this application according to specific circumstances.

[0030] In one embodiment, if Figure 1 As shown, the first aspect of the present invention provides a distributed collaborative optimization control method for a distribution network, comprising:

[0031] S1. Real-time monitoring of distribution network loss data;

[0032] In one embodiment, before the real-time monitoring of the power distribution network loss data, the method further includes:

[0033] The distribution network is divided into regions by using a stable connection area in the distribution network as a grid edge to obtain a plurality of edge areas; the stable connection area is configured to be statically connected within the area, the areas are not directly connected to each other, the power supply of the area is the upper-level substation, and the boundary between adjacent areas is a normally open tie switch in the distribution network;

[0034] Modeling the electrical equipment in each of the edge areas to describe the equipment parameters, operating status, installation location, and control or adjustment methods of each edge area to generate a basic attribute model;

[0035] An edge agent is configured for each edge area, and each basic attribute model is used as the static structure of each edge agent. Real-time operation data in each edge area is collected and stored through each basic attribute model to construct the operation status of each edge agent.

[0036] Specifically, before real-time monitoring of the distribution network loss data, the present invention needs to divide the distribution network into multiple edge areas, and configure a distribution network edge agent for each edge area as the edge computing carrier of the optimization strategy to perform data collection and processing, optimization control model establishment and optimization solution calculation, and control instruction issuance tasks for controllable resource devices within the edge area under its jurisdiction. The area division process is as follows:

[0037] First, the radial source-grid-load-storage distribution network to which the present invention is applicable is connected to the large power grid, and is composed of a 10kV busbar of a substation, a 10kV distribution feeder, a distribution transformer, a switch or ring network cabinet, a pole-mounted switch, a cable branch box, a switchgear and other power equipment, as well as users, substations, photovoltaics, energy storage, etc. connected to the distribution network; then, the stable connection area of ​​the distribution network is divided into edges as the edge distribution network of the power grid. The stable connection area of ​​the distribution network can be generated according to the static topology of the distribution network, and the following conditions are met: ① static connectivity within the area; ② the areas are not directly connected, that is, not directly connected through 10 kV lines; ③ the power supply of the area is the upper-level substation, that is, 220 / 110 / 35 kV substation, the boundary between adjacent distribution network edge agents is the normally open interconnecting switch in the distribution network; then, the electrical equipment in each demarcated edge area is modeled to describe the basic properties of the equipment, such as equipment parameters, operating status, installation location, and control or adjustment method, and this information is used as the static structure of the edge agent in the corresponding area, becoming the benchmark data of the agent state, and generating the basic attribute model of each edge area; finally, the real-time operation data and real-time network loss data in the corresponding edge area are collected through each edge agent, including equipment output, line power, line current, node injection power, node voltage value, and collective data of load, interconnecting switch and line, and these data are stored according to the established basic attribute model to be used to construct the operating status of the agent, providing data support for subsequent event judgment and optimization.

[0038] The present invention divides the distribution network into several edge areas, with each area serving as a management unit, which helps to achieve regional management of the distribution network, making the operation and maintenance of the distribution network more refined and improving management efficiency. Edge intelligent agents are configured for each edge area. These intelligent agents can monitor the status of electrical equipment in the area in real time, detect and handle abnormal situations in a timely manner, thereby improving the reliability and safety of the distribution network. By introducing edge intelligent agents and basic attribute models, the distribution network can achieve real-time monitoring, early warning, control and optimization functions, thereby improving the intelligence level of the distribution network, helping to reduce the operation and maintenance costs of the distribution network, improving energy utilization efficiency, and promoting the sustainable development of the power system.

[0039] S2. When the network loss data exceeds a preset network loss threshold, the network active power loss of the distribution network is used as an objective function, the line flow, distributed power generation power, and node voltage in the distribution network are constrained respectively, and the resulting constraints are reconstructed through semidefinite programming. An optimization control model of the distribution network is constructed based on the reconstructed results and the objective function.

[0040] Under normal circumstances, the distribution network operates according to the set planned working mode, but considering the economic efficiency of the distribution network system operation, the present invention uses system network loss as a condition for the edge intelligent agent to trigger the action. When each edge intelligent agent monitors the event of excessive system network loss, it starts the optimization business processing process, and optimizes the reactive power of the distribution network system by adjusting the distributed power supply and other equipment in the system, thereby reducing the network loss and stabilizing the node voltage; wherein, the system network loss refers to the electric energy loss caused by factors such as the resistance, inductance and capacitance of power equipment such as lines, transformers and switchgear during the transmission and distribution process in the distribution network; the system's preset network loss threshold refers to a threshold set for controlling and monitoring the system network loss, which can be determined based on factors such as system capacity, load demand, and economic environment.

[0041] The present invention monitors the network loss data of the distribution network system in real time through each edge intelligent agent. When the network loss data exceeds the preset network loss threshold, a distribution network excessive loss event is triggered. An optimization model needs to be constructed based on the data at this time, so that each edge intelligent agent can take optimization control behavior to solve the model, and adjust the line flow, distributed power supply power and node voltage in the area according to the optimization strategy obtained by the solution to ensure that the distribution network operates in the optimal state, thereby eliminating the event and returning the distribution network to a satisfactory operating state.

[0042] The present invention takes the network active power loss of the distribution network as the objective function, which is expressed by the following formula:

[0043]

[0044] Where, is the network active power loss of the distribution network; n is the total number of nodes; lij is the square value of the current amplitude of line ij; r ij is the resistance of circuit ij; P ij , Q ij are respectively the active and reactive power transmitted at the head end of line ij; V i is the voltage at node i.

[0045] Constraints are imposed on the line flow, distributed generation power, and node voltage in the distribution network to form constraint conditions. The line flow constraint is an equality constraint, and the branch flow equation is used. For radial source-grid-load-storage distribution networks, the effects of line-to-ground capacitance and susceptance are ignored. According to Kirchhoff's voltage-current law, the flow equation described in branch form can be obtained as follows:

[0046]

[0047] Where, P jk , Q jk are the active and reactive powers transmitted at the head end of line jk, respectively; k:j→k represents the set of all lines with node j as the head end and node k as the end end; 、 are the active and reactive power generated by the distributed generation at node j; x ij is the reactance of line ij; p c j ,q c j are the active and reactive power of the load at node j respectively; is the voltage amplitude at node j; 、V ref are the equilibrium node voltage amplitude and its reference value respectively.

[0048] The reactive power of the controllable distributed power equipment in the basic attribute model is the power constraint of the distributed power supply. The adjustment range is limited by its own installation capacity and active power, which is expressed by the following formula:

[0049]

[0050] Where s j is the installed capacity of distributed generation equipment; q gmax j is the maximum reactive power of the distributed generation device; p g j It is the current operating value of the active power of the distributed generation device.

[0051] At the same time, considering the stability and reliability of the distribution network operation, it is required that the voltage of each node in the distribution network meets the safety constraints and remains near the rated voltage. That is, the node voltage constraint is expressed by the following formula:

[0052]

[0053] Where, It is the voltage deviation, usually 0.05 pu.

[0054] It can be seen from this that the distributed power constraints and node voltage constraints are non-equality constraints, and the line flow constraints contain complex quadratic terms, which makes the optimization problem non-convex. It is difficult for distributed optimization algorithms to find the global optimal solution for non-convex problems and convergence cannot be guaranteed. In order to solve non-convex nonlinear programming problems in a distributed manner, the branch flow equation is first reasonably simplified, that is, 1) the power loss on the line is much smaller than the power transmitted by the line itself; 2) the voltage deviation between nodes is much smaller than the node voltage itself. Based on the above two assumptions, the quadratic terms in the branch flow equation can be ignored, and the voltage amplitude V of each node in the objective function can also be approximated as the voltage amplitude V1 of the equilibrium node; then the constraints are semi-definitely reconstructed, thereby converting the original non-convex optimization problem into a convex optimization problem. To further ensure the convexity of the flow constraints and the safety of the node voltage, the semi-definite programming reconstruction of the resulting constraints is performed using the following formula:

[0055]

[0056] Where U i is the square value of the voltage amplitude at node i; U j is the square value of the voltage amplitude at node j; is the norm.

[0057] The optimal control model of the distribution network obtained is expressed as follows:

[0058]

[0059] Where q g Reactive power generated by distributed power sources; is the square value of the voltage amplitude of the equilibrium node; N is the set of nodes in the distribution network; 、 are the minimum and maximum values ​​of reactive power generated by the distributed generation at node j.

[0060] After reasonable simplification and reconstruction, the original optimization problem is transformed into a convex quadratic programming model with a quadratic objective function and linear constraints, that is, an optimization control model. According to optimization theory, the local optimal solution of a convex optimization problem is equal to the global optimal solution. This makes it easy to find the global optimal solution to this optimization problem using a distributed algorithm, thereby effectively improving the operational stability of the distribution network.

[0061] S3. Acquire coupling branch states between a plurality of edge areas in the distribution network, and decompose the optimization control model into a plurality of distributed optimization control sub-models based on partition coordination based on the coupling branch states;

[0062] In one embodiment, step S3 includes:

[0063] Obtaining a node set and a branch set in each edge region to couple adjacent edge regions through branches on edge boundaries to obtain coupling branches between the edge regions, and using the power transmitted on each coupling branch and the square of the voltage at the nodes at both ends of each branch as the corresponding coupling branch state;

[0064] Based on the principle that the coupling branch state of each edge area is equal to the coupling branch state of its adjacent edge area, the optimization control model is decomposed into a number of distributed optimization control sub-models based on partition coordination; the number of the distributed optimization control sub-models is consistent with the number of the edge areas.

[0065] Specifically, for a distribution network system, the graph structure can be used Indicates that N is the set of nodes in the entire system, E is the set of branches in the entire system, and e ij ∈E represents a branch in the system, where Assume that the distribution network system is divided into r edge areas, represented by set R; for a certain edge area a, N a Represents the node set of the area, E a represents the branch set in the area, e ij ∈E a Represents a branch in the region. Adjacent edge regions are coupled through branches on the edge boundaries, that is, two or more edge regions that share at least one branch are adjacent edge regions. The set of coupled branches is denoted by Indicates. Coupling branch e ij The state variables include the power P transmitted by the branch ij , Q ij , the voltage square of the nodes at both ends of the branch U i 、U j Then the state of the coupling branch in the edge region a is represented by the vector The schematic diagram of the distribution network edge area division is as follows: Figure 2 As shown, with the attached Figure 2 Take the distribution network system shown in the figure as an example. The node set of edge area 1 is {1,2,3,4,5}, the node set of edge area 2 is {4,5,6,7,8}, and the node set of edge area 3 is {7,8,9,10,11}. Then edge area 1 and edge area 2 are connected through branch e. 45 To couple, edge area 2 and edge area 3 are connected through branch e 78 Coupling, the collection of coupling branches .

[0066] In order to make the problem after edge region division equivalent to the original problem, the coupled branch state variables obtained by solving the adjacent edge region subproblems must be equal, that is, the coupled branch state variables obtained by the edge region a subproblem are a,ij The coupled branch state X obtained from the subproblem b in the adjacent edge region b,ij must be equal. Based on this, the optimization control model can be decomposed into several distributed optimization control sub-models based on partition coordination. The number of these sub-models is equal to the number of edge regions and edge agents. Each distributed optimization control sub-model is expressed by the following formula:

[0067]

[0068]

[0069] Where r is the total number of edge regions; f a (x a ),h a (x a ), g a (x a ) are the objective function, equality constraint, and inequality constraint of the distributed optimization control sub-model of the a-th edge region respectively; x a is the decision variable of the ath edge region, that is, the coupling branch state ; b is other edge regions adjacent to the a-th edge region, so b does not refer to a fixed region, but all edge regions adjacent to edge region a.

[0070] At this point, the optimization control model of the entire system is decomposed into the collaborative calculation of r distributed optimization control sub-models, and a distributed collaborative optimization control model of multiple distribution network edge intelligent agents is constructed. By obtaining the node set and branch set of each edge area, the present invention helps to clearly define the boundaries and internal structure of each edge area, thereby improving the zoning management capability of the distribution network. Through clear regional division, resource allocation, fault location and isolation, and the formulation of operation strategies can be more effectively carried out; by determining the coupling branches between adjacent edge areas, electrical connection and information interaction between areas are achieved, which helps to enhance the coordination and control capabilities between them while maintaining the relative independence of each edge area. When a fault occurs in a certain area or the operation strategy needs to be adjusted, information can be quickly transmitted through the coupling branches to achieve rapid response and coordinated action between areas; based on the principle that the coupling branch state of each edge area is equal to the coupling branch state of its adjacent edge area, the optimization control model is decomposed into several distributed optimization control sub-models based on partition coordination, which reduces the complexity of the control model and enables each edge area to independently solve its own optimization control sub-model, thereby improving the efficiency and scalability of the calculation. At the same time, the distributed implementation method also helps to improve the robustness and fault tolerance of the system.

[0071] S4. Controlling the edge agents deployed in each of the edge areas to solve each of the distributed optimization control sub-models using a preset distributed optimization algorithm to obtain a distributed collaborative optimization strategy; the preset distributed optimization algorithm is a synchronous alternating direction multiplier algorithm that introduces historical iteration results and increments of each of the edge agents;

[0072] In one embodiment, step S4 includes:

[0073] Based on each of the distributed optimization control sub-models, control each of the edge agents to construct an augmented Lagrangian objective function and edge constraints for the edge region to which it belongs;

[0074] Under the edge constraints, each edge agent is controlled to use the preset distributed optimization algorithm to iteratively solve the augmented Lagrangian objective function until the preset iterative convergence condition is reached, and the regional decision variables of each edge area are output as a distributed collaborative optimization strategy.

[0075] Specifically, since the decomposed distributed optimization control sub-model is a convex programming model with a separable objective function and linear boundary coupling constraints, the present invention adopts a synchronous alternating direction multiplier algorithm that introduces the historical iteration results of the edge agent and the coupling branch state increment to achieve distributed solution. Still taking the edge area a and its adjacent edge area b as an example, the distributed optimization control sub-models corresponding to these two edge areas are the edge a sub-problem and the edge b sub-problem (wherein the edge b sub-problem corresponds to the distributed optimization control sub-model of all adjacent edge areas of the edge area a), and then based on these two sub-problems, the augmented Lagrangian objective function corresponding to the objective function of each edge agent is constructed. and , and through appropriate transformations, it is transformed into:

[0076]

[0077]

[0078] Where t is the number of iterations; 、 are the initial values ​​of the coupling branch states in edge regions a and b, respectively; and are the fixed reference values ​​of the edge regions a and b at the t+1th iteration respectively; a and λ b are the dual variables of edge regions a and b, each containing 4 elements, such as Corresponding to the coupling branch status ; is the penalty parameter; is the square of the 2-norm.

[0079] Next, the edge constraints of each edge region are determined, where the edge constraint of edge region a can be expressed as:

[0080]

[0081] That is, the constraints in the optimization control model of the aforementioned distribution network are limited to the edge area a. Similarly, the edge constraints of the edge area b also limit the constraints in the optimization control model of the aforementioned distribution network to its corresponding edge area. I will not go into details here.

[0082] For the solution of each distributed optimization control sub-model, that is, under the edge constraints corresponding to each edge region, with the goal of minimizing the augmented Lagrangian objective function, the edge agent corresponding to the region adopts the synchronous alternating direction multiplier algorithm that introduces the historical iteration results of the edge agent and the coupling branch state increment to iteratively solve the augmented Lagrangian objective function of the region until the preset iterative convergence condition is reached. The regional decision variables of each edge region (that is, the last updated coupling branch state) are output and used as a distributed collaborative optimization strategy to control each edge agent to adjust the line flow, distributed power supply power and node voltage in the region to ensure that the distribution network operates in the optimal state. Among them, the preset iterative convergence condition is that the boundary residual tends to zero. The boundary residual is the square of the two norms of the difference between the coupling branch states obtained in adjacent edge regions, as shown in the following formula:

[0083]

[0084]

[0085] Where, is the convergence accuracy.

[0086] When the regional decision variables output by each edge region satisfy the above formula, the iteration ends. The present invention decomposes the large-scale optimization problem into multiple sub-problems through distributed optimization control, which are processed in parallel by each edge agent, thereby significantly improving the optimization efficiency; the distributed optimization strategy makes the system more tolerant to the failure or failure of a single agent, and enhances the robustness and reliability of the system; by constructing an augmented Lagrangian objective function and considering edge constraints, the problem can be described and optimized more accurately, and the rational allocation and efficient utilization of resources can be achieved; each edge agent works together under the distributed optimization control framework to jointly solve the optimization problem, which helps to achieve a global optimal solution or an approximate optimal solution.

[0087] In another embodiment, the solution of each distributed optimization control sub-model can be achieved by combining a distributed stochastic gradient descent algorithm with a federated learning method: still taking the edge area a and its adjacent edge area b as an example, the objectives and coupling constraints are determined according to the distributed optimization control sub-models of edge area a and edge area b; 1. Initialization phase: initialize the global model parameters, and distribute the global model parameters to edge agents a and b, set the learning rate, communication cycle and maximum number of iterations; 2. Local update phase: Stochastic gradient descent update: each edge agent randomly samples a mini-batch from the local data set, calculates the stochastic gradient to update the local variables, local iteration: in each communication cycle, a and b respectively perform the stochastic gradient descent update process for the communication cycle times; 3. Global aggregation: at the end of each communication cycle, a and b upload the local model parameters to the central server for global aggregation. The central server receives the model parameters of a and b to calculate the global model parameters and distribute them to a and b; 4. Check convergence: a and b calculate the boundary residual based on the obtained parameters. If the convergence conditions are met, the algorithm terminates and outputs the coupling branch states of their respective regions; otherwise, continue to iterate the above process. This solution can accelerate model training, protect data privacy, improve model generalization capabilities, has flexibility and scalability, and promotes efficient resource utilization. It is particularly suitable for processing large-scale data sets and distributed computing environments, and provides new ideas and solutions for the training and deployment of machine learning models.

[0088] In one embodiment, the controlling each of the edge agents to iteratively solve each of the augmented Lagrangian objective functions using the preset distributed optimization algorithm includes:

[0089] Initializing the dual variables and penalty parameters of each edge agent, and initializing the coupling branch state and fixed reference value of each edge agent according to the historical iteration results and coupling branch state increments of each edge agent;

[0090] Solving the augmented Lagrangian objective function of each edge agent using a gradient descent method according to the initialized dual variables, penalty parameters, and fixed reference values ​​to obtain regional decision variables of each edge region;

[0091] When the regional decision variable does not reach the preset iterative convergence condition, controlling each edge agent to interact with its own regional decision variable with its adjacent edge agent, so that each edge agent updates its own fixed reference value and dual variable according to the regional decision variable of each adjacent edge agent;

[0092] Updating the penalty parameter by an adaptive penalty parameter adjustment method so that each edge agent solves each augmented Lagrangian objective function based on the updated fixed reference value, dual variable, and penalty parameter to obtain updated regional decision variables;

[0093] When the updated regional decision variables do not meet the preset iterative convergence conditions, the updating steps of fixing the reference value, penalty parameter and dual variable and the solving steps of the augmented Lagrangian objective function are repeatedly executed based on the updated regional decision variables until the updated regional decision variables meet the preset iterative convergence conditions.

[0094] Specifically, the various parameters required in the synchronous alternating direction multiplier algorithm are first initialized, that is, the initial dual variables and penalty parameters are set for each edge agent. These parameters will be continuously updated in the subsequent iteration process. Among them, considering that the edge agent can store the previous iteration value, the present invention pre-updates the decision variable value by comprehensively utilizing the historical iteration results and the coupling branch state increment, thereby improving the convergence performance. Then, the historical iteration results and increments of each edge agent are used to update its own coupling branch state as follows:

[0095]

[0096]

[0097]

[0098]

[0099]

[0100]

[0101] Where, 、 are the state increments of the coupling branches in the edge regions a and b respectively; t is the number of iterations; 、 They are the coupling branch states after initialization of edge regions a and b respectively; The number of historical iteration values ​​stored for the edge agent; 、 are the incremental coefficients of edge regions a and b, respectively, which are determined by solving the minimization error; 、 are the historical increment coefficients corresponding to the edge areas a and b at the i-th historical iteration.

[0102] At the same time, the historical iteration results of each edge agent are used to initialize the fixed reference value of each edge agent, which is expressed by the following formula:

[0103]

[0104] Then, the gradient descent method is used to solve the augmented Lagrangian objective function of each edge region based on the initialized dual variables, penalty parameters and fixed reference values ​​to obtain the initial regional decision variables, namely:

[0105]

[0106]

[0107] Based on the aforementioned boundary residual and the preset number of iterations, the initial regional decision variables are judged to see whether they meet the conditions. If not, the regional decision variables of each edge agent and its adjacent edge agent are controlled to interact with each other (that is, the coupling branch state obtained by the above formula). and ), and use this to calculate the average value of the coupling branch state of each edge agent as a fixed reference value for the next iteration:

[0108]

[0109] Control each edge agent to update the dual variables of the region:

[0110]

[0111]

[0112] The adaptive penalty parameter adjustment method is used to update the penalty parameter, which is expressed by the following formula:

[0113]

[0114] Where, 、 are the penalty parameters before and after the update respectively; and It is an adjustable parameter and can be set to v = 2. is 10; is the fixed reference value at the tth iteration; is the 2-norm, also known as the Euclidean norm. Since the convergence of the synchronous alternating direction multiplier algorithm is related to the globality of the solution and the penalty parameter, an adaptive penalty parameter adjustment method can be used to maintain the globality of the solution while meeting convergence requirements.

[0115] Then, each edge agent is controlled to solve each augmented Lagrangian objective function based on the updated fixed reference value, dual variable and penalty parameter by gradient descent method or other algorithms to obtain the updated regional decision variables; continue to judge whether the updated regional decision variables converge. If not, repeat the updating steps of fixed reference value, penalty parameter and dual variable and the solving steps of each augmented Lagrangian objective function based on the updated regional decision variables until the updated regional decision variables reach the preset iterative convergence condition. The solution process of each distributed optimization control sub-model is as follows: Figure 3 shown.

[0116] Through the collaboration of edge intelligent agents, the present invention realizes the distributed optimization of regional decision variables, thereby improving computational efficiency and scalability. Since data does not need to be centrally processed, the data privacy of edge intelligent agents is protected. Through the adaptive penalty parameter adjustment method, the scheme can dynamically adjust the penalty parameter according to actual conditions, thereby improving the convergence speed and optimization effect. By presetting iterative convergence conditions, the scheme can ensure that the final regional decision variables meet certain accuracy requirements.

[0117] In one embodiment, the IEEE 33-node system topology and edge division diagram is as follows: Figure 4 As shown, with the attached Figure 4 The effectiveness of the proposed method is verified in the distribution network shown in the figure. The IEEE 33-node is divided into three edge regions, each of which is managed by an edge agent. According to the partitioning method of copying the boundary branches and placing them in two adjacent edge regions at the same time, the node sets are represented as edge region 1: {1-8, 19-26}, edge region 2: {7-18}, and edge region 3: {6, 26-33}. The locations of the five controllable distributed generation units are nodes {8, 12, 16, 21, 30}. The active output of each distributed generation unit is set to 400 kW, the maximum reactive output is set to 400 kvar, and the voltage safety constraint upper limit is 1.05 pu, and the lower limit is 0.95 pu.

[0118] The power flow calculation uses per-unit values, with the system's base power set at 1.0 MVA, the voltage base selected at 12.66 kV, and the balancing node at node 1 at a voltage of 1.0 pu. At this point, the voltage deviation at terminal node 33 in the IEEE 33-node system reaches 0.052 pu, resulting in a network loss of 202.7 kW. This triggers a high distribution network loss event (assuming the edge agent sets the permissible loss threshold at 180 kW). The edge agent must take action to resolve the event.

[0119] The distributed collaborative optimization control method of the distribution network described in the present invention is used to optimize the distributed power supply in the system, with the penalty parameter taken as 0.05 and the convergence accuracy taken as , we can get the voltage comparison diagram of IEEE 33-bus system before and after optimization as shown in the figure below: Figure 5 As shown in the attached Figure 5 It can be seen that after optimized control, the voltage distribution of the three edge systems has been significantly improved, the voltage of all nodes meets the requirements for safe operation, and the voltage deviation is significantly reduced. After optimized control, the network loss of the IEEE 33-node system is reduced to 36.454kW, which is within the allowable threshold set by the edge intelligent agent, eliminating the excessive loss event in the distribution network and improving the economic efficiency of the distribution network operation.

[0120] In order to verify the correctness and effectiveness of the distributed collaborative optimization control method for the distribution network described in the present invention, its control effect is compared with that of the centralized optimization control method. Among them, the centralized optimization method uses the primal-dual interior point method for solution. The reactive output of the system distributed power supply calculated by the distributed method and the centralized method is compared as shown in the following table:

[0121]

[0122] It can be seen from the above table that the output of distributed power sources under distributed optimization control is the same as that under centralized optimization control, which shows the correctness of the results obtained by using the improved synchronous alternating direction multiplier algorithm to distribute the optimization control model.

[0123] Distributed solution is an iterative optimization process that requires each edge agent to exchange the state information of the coupled branch in each iteration. Therefore, the fewer iterations required for solution convergence, the fewer communications are required to obtain a converged collaborative control strategy, and the smaller the communication burden. The distributed collaborative optimization control method proposed in this invention is applied to the conventional ADMM algorithm to obtain the boundary residual iterative change diagram of each edge as shown in the figure below. Figure 6 As shown in FIG, the method proposed in the present invention converges after 13 iterations, while the conventional ADMM algorithm requires 20 iterations to complete the convergence. This shows that the method proposed in the present invention has a faster convergence speed.

[0124] The iterative change diagram of the reactive power output of the distributed power generation of the IEEE 33-bus system is as follows: Figure 7 As shown in the figure, the reactive output of the distributed power generation system obtained from the optimization of the three edges changes with the iterative process. It can be seen that the distributed optimization control method described in this invention can quickly converge on all three edges, requiring fewer iterations, reducing the communication burden between adjacent edge agents, and quickly achieving the control strategy for their respective edge regions. Because the optimization subproblems established by each edge agent are simple convex quadratic programming problems, each iteration yields a global optimal solution, facilitating rapid convergence to a consistent control strategy for each region. Furthermore, each edge agent can quickly solve the suboptimization problem.

[0125] In order to analyze the adaptability of the distributed method proposed in this invention to the changes in the system operating state, the daily load curve of the IEEE33-node system is shown as follows: Figure 8 As shown, the power of each load is as follows Figure 8 The 96-point normalized daily load curve shown in Figure 2 shows the changes in the load curve. The active power values ​​of each load in the previous analysis correspond to the following Figure 8 The load value corresponding to the normalized load at the 68th moment is 1.0, and the load power factor at each moment remains unchanged. Distributed optimization control is performed under the system state at each moment, and the network loss comparison before and after optimization of the IEEE 33-node system is obtained as shown in the attached figure. Figure 9 As shown in the attached Figure 9 It can be seen that in the three edge subsystems, as the load continues to change, the edge agent autonomy and multi-party collaboration method can achieve distributed optimization control of the system at every moment, significantly reducing network loss and achieving the business goal of network loss optimization. It also shows that the proposed method has good adaptability to changes in system operating status.

[0126] In the embodiment of the present application, based on the problem of how to effectively improve the operational stability of the distribution network, a distributed collaborative optimization control method for the distribution network is designed. By dividing the global distribution network into multiple areas and constructing a distribution network edge intelligent agent as an edge computing carrier of the optimization strategy, the local execution of data collection, processing and optimization control is realized, which not only improves the data processing efficiency, but also reduces the dependence on the central control system and enhances the reliability and flexibility of the system; when the monitoring system detects that the network loss is too high, the edge intelligent agent can quickly start the network loss optimization business processing process, and effectively reduce the system network loss by adjusting distributed power supplies and other equipment, thereby improving the node voltage stability. This real-time, automated optimization control strategy helps to improve the safe and economical operation level of the power system and reduce energy waste; through reasonable simplification, convexification and iterative processing, the non-convex optimization problem is converted into a convex optimization problem to construct an optimization control model, which ensures the convergence and global optimality of the distributed algorithm solution, not only simplifies the solution process of the optimization problem, but also improves the accuracy and reliability of the optimization results; in terms of distributed collaborative control, the present invention realizes information sharing and collaborative optimization between adjacent edge intelligent agents by constructing multiple distributed collaborative optimization sub-models, which helps to balance the power supply and demand in various regions and improve the stability and economy of the entire power system.

[0127] It should be noted that although the steps in the above flowchart are shown in sequence as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders.

[0128] In another embodiment, if Figure 10 As shown, the second aspect of the present invention provides a distributed collaborative optimization control system for a distribution network, comprising:

[0129] The network loss monitoring module 10 is used to monitor the network loss data of the distribution network in real time;

[0130] a model construction module 20 for, when the network loss data exceeds a preset network loss threshold, using the network active power loss of the distribution network as an objective function, respectively constraining the line flow, distributed power generation power, and node voltage in the distribution network, and performing semi-definite programming reconstruction on the resulting constraints, and constructing an optimization control model for the distribution network using the reconstruction results and the objective function;

[0131] A model decomposition module 30 is configured to obtain coupling branch states between a plurality of edge regions in the distribution network, and decompose the optimization control model into a plurality of distributed optimization control sub-models based on partition coordination based on the coupling branch states;

[0132] The strategy solving module 40 is used to control the edge intelligent agents deployed in each of the edge areas to use a preset distributed optimization algorithm to solve each of the distributed optimization control sub-models to obtain a distributed collaborative optimization strategy; the preset distributed optimization algorithm is a synchronous alternating direction multiplier algorithm that introduces the historical iteration results and increments of each of the edge intelligent agents.

[0133] It should be noted that each module in the above-mentioned distributed collaborative optimization control system of a distribution network can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules. For the specific definition of a distributed collaborative optimization control system of a distribution network, please refer to the definition of a distributed collaborative optimization control method of a distribution network above. The two have the same functions and effects and will not be repeated here.

[0134] A third aspect of the present invention provides an electronic device, comprising:

[0135] processor, memory, and bus;

[0136] The bus is used to connect the processor and the memory;

[0137] The memory is used to store operation instructions;

[0138] The processor is used to call the operation instruction, and the executable instruction enables the processor to perform operations corresponding to the distributed collaborative optimization control method of the distribution network as shown in the first aspect of the present application.

[0139] In an alternative embodiment, an electronic device is provided, such as Figure 11 As shown, Figure 11 The electronic device 5000 shown includes: a processor 5001 and a memory 5003. The processor 5001 and the memory 5003 are connected, for example, via a bus 5002. Optionally, the electronic device 5000 may further include a transceiver 5004. It should be noted that in actual applications, the number of transceivers 5004 is not limited to one, and the structure of the electronic device 5000 does not constitute a limitation on the embodiments of the present application.

[0140] Processor 5001 may be a CPU, a general-purpose processor, a DSP, an ASIC, an FPGA, or other programmable logic device, a transistor logic device, a hardware component, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 5001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.

[0141] The bus 5002 may include a path for transmitting information between the above components. The bus 5002 may be a PCI bus or an EISA bus, etc. The bus 5002 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 11 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0142] The memory 5003 may be a ROM or other type of static storage device that can store static information and instructions, a RAM or other type of dynamic storage device that can store information and instructions, or an EEPROM, a CD-ROM or other optical disk storage, an optical disc storage (including a compact disc, a laser disc, an optical disc, a digital versatile disc, a Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to these.

[0143] The memory 5003 is used to store application code for executing the solution of the present application, and the execution is controlled by the processor 5001. The processor 5001 is used to execute the application code stored in the memory 5003 to implement the content shown in any of the above method embodiments.

[0144] Among them, electronic devices include but are not limited to: mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc.

[0145] The fourth aspect of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the distributed collaborative optimization control method of the distribution network shown in the first aspect of the present application.

[0146] Another embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon. When the computer-readable storage medium is run on a computer, the computer can execute the corresponding contents of the aforementioned method embodiments.

[0147] In addition, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above method when executed by a processor.

[0148] In summary, the present invention relates to the field of intelligent optimization operation of distribution networks, and discloses a distributed collaborative optimization control method, system, equipment and medium for distribution networks. The method partitions the distribution network and configures edge intelligent agents to monitor the network loss data of the distribution network in real time. When the network loss exceeds a preset threshold, the optimization control process is triggered: with the network active loss as the objective function, the line flow, distributed power supply power and node voltage are constrained and reconstructed through semi-definite programming to construct an optimization control model as a convex quadratic programming model; based on the coupling branch states between each area, the constructed model is decomposed into several distributed optimization control sub-models based on partition coordination, and a synchronous alternating direction multiplier algorithm that introduces the historical iteration results and increments of each edge intelligent agent is used to solve each sub-model to obtain a distributed collaborative optimization strategy and execute it; through partition management, distributed optimization and collaborative control, efficient operation and optimization of the distribution network are achieved.

[0149] Each embodiment in this specification is described in a progressive manner, and the same or similar parts of each embodiment can be directly referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be noted that the various technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0150] The above-described embodiments merely represent several preferred implementations of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent. It should be noted that a person skilled in the art could make several improvements and substitutions without departing from the technical principles of the present invention, and these improvements and substitutions should also be considered within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be based on the scope of protection of the claims.

Claims

1. A distributed collaborative optimization control method for a distribution network, characterized in that: include: Real-time monitoring of distribution network loss data; When the network loss data exceeds a preset network loss threshold, the network active power loss of the distribution network is used as an objective function, the line flow, distributed power source power and node voltage in the distribution network are constrained respectively, and the resulting constraint conditions are reconstructed by semi-definite programming. The optimization control model of the distribution network is constructed based on the reconstruction results and the objective function; Acquiring coupling branch states between a plurality of edge areas in the distribution network, and decomposing the optimization control model into a plurality of distributed optimization control sub-models based on partition coordination based on the coupling branch states; Controlling the edge agents deployed in each of the edge areas to solve each of the distributed optimization control sub-models using a preset distributed optimization algorithm to obtain a distributed collaborative optimization strategy; The preset distributed optimization algorithm is a synchronous alternating direction multiplier algorithm that introduces the historical iteration results and increments of each edge agent; The control of the edge agents deployed in each edge area uses a preset distributed optimization algorithm to solve each distributed optimization control sub-model to obtain a distributed collaborative optimization strategy, including: Based on each of the distributed optimization control sub-models, control each of the edge agents to construct an augmented Lagrangian objective function and edge constraints for the edge region to which it belongs; Under the edge constraints, control each edge agent to iteratively solve each augmented Lagrangian objective function using the preset distributed optimization algorithm until a preset iterative convergence condition is reached, and output the regional decision variables of each edge region as a distributed collaborative optimization strategy; The controlling each of the edge agents to iteratively solve each of the augmented Lagrangian objective functions using the preset distributed optimization algorithm includes: Initializing the dual variables and penalty parameters of each edge agent, and initializing the coupling branch state and fixed reference value of each edge agent according to the historical iteration results and coupling branch state increments of each edge agent; Solving the augmented Lagrangian objective function of each edge agent using a gradient descent method according to the initialized dual variables, penalty parameters, and fixed reference values ​​to obtain regional decision variables of each edge region; When the regional decision variable does not reach the preset iterative convergence condition, controlling each edge agent to interact with its own regional decision variable with its adjacent edge agent, so that each edge agent updates its own fixed reference value and dual variable according to the regional decision variable of each adjacent edge agent; Updating the penalty parameter by an adaptive penalty parameter adjustment method so that each edge agent solves each augmented Lagrangian objective function based on the updated fixed reference value, dual variable, and penalty parameter to obtain updated regional decision variables; When the updated regional decision variables do not meet the preset iterative convergence condition, repeatedly performing the steps of updating the fixed reference value, the penalty parameter, and the dual variable, and solving the augmented Lagrangian objective function based on the updated regional decision variables until the updated regional decision variables meet the preset iterative convergence condition; The historical iteration results and incremental updates of the edge agents’ own coupling branch states are expressed as follows: Where, 、 are the state increments of the coupling branches in the edge regions a and b respectively; t is the number of iterations; 、 They are the coupling branch states after initialization of edge regions a and b respectively; 、 are the initial values ​​of the coupling branch states of edge regions a and b respectively; m is the number of historical iteration values ​​stored by the edge agent; 、 are the incremental coefficients of edge regions a and b respectively; 、 are the historical increment coefficients corresponding to the edge regions a and b at the i-th historical iteration respectively; The adaptive penalty parameter adjustment method is expressed by the following formula: Where, 、 are the penalty parameters before and after the update respectively; and is an adjustable parameter; is the fixed reference value at the tth iteration.

2. A distributed collaborative optimization control method for a distribution network according to claim 1, characterized in that: Before the real-time monitoring of the power distribution network loss data, the method further includes: The distribution network is divided into regions by using a stable connection area in the distribution network as a grid edge to obtain a plurality of edge areas; the stable connection area is configured to be statically connected within the area, the areas are not directly connected to each other, the power supply of the area is the upper-level substation, and the boundary between adjacent areas is a normally open tie switch in the distribution network; Modeling the electrical equipment in each of the edge areas to describe the equipment parameters, operating status, installation location, and control or adjustment methods of each edge area to generate a basic attribute model; An edge agent is configured for each edge area, and each basic attribute model is used as the static structure of each edge agent. Real-time operation data in each edge area is collected and stored through each basic attribute model to construct the operation status of each edge agent.

3. The distributed collaborative optimization control method for a distribution network according to claim 1, characterized in that: The semidefinite programming reconstruction of the formed constraints is performed by the following formula: Where, P ij , Q ij are respectively the active and reactive power transmitted at the head end of line ij; l ij is the square value of the current amplitude of line ij; U i is the square value of the voltage amplitude at node i; is the voltage amplitude at node j; U j is the square value of the voltage amplitude at node j; The optimal control model of the distribution network is expressed by the following formula: Where q g is the reactive power generated by the distributed generation; n is the total number of nodes; rij and xij are the resistance and reactance of line ij respectively; is the square value of the equilibrium node voltage amplitude; P jk , Q jk are the active and reactive power transmitted at the head end of line jk, respectively; k:j→k represents the set of all lines with node j as the head end and node k as the end end; N is the set of nodes in the distribution network; p c j ,q c j are the active and reactive power of the load at node j respectively; p g j is the active power generated by the distributed power supply at node j; V ref is the reference value of the equilibrium node voltage amplitude; 、 are the minimum and maximum values ​​of reactive power generated by the distributed generation at node j; is the voltage deviation.

4. The distributed collaborative optimization control method for a distribution network according to claim 1, characterized in that: The obtaining of coupling branch states between a plurality of edge areas in the distribution network and decomposing the optimization control model into a plurality of distributed optimization control sub-models based on partition coordination based on the coupling branch states includes: Obtaining a node set and a branch set in each edge region to couple adjacent edge regions through branches on edge boundaries to obtain coupling branches between the edge regions, and using the power transmitted on each coupling branch and the square of the voltage at the nodes at both ends of each branch as the corresponding coupling branch state; Based on the principle that the coupling branch state of each edge region is equal to the coupling branch state of its adjacent edge region, the optimization control model is decomposed into a number of distributed optimization control sub-models based on partition coordination; the number of the distributed optimization control sub-models is consistent with the number of the edge regions; the distributed optimization control sub-models are expressed by the following formula: Where r is the total number of edge regions; f a (x a ),h a (x a ), g a (x a ) are the objective function, equality constraint, and inequality constraint of the distributed optimization control sub-model of the a-th edge region respectively; x a is the coupling branch state of the a-th edge region; b is other edge regions adjacent to the a-th edge region.

5. A distributed collaborative optimization control system for a distribution network, characterized in that: include: Network loss monitoring module, used to monitor network loss data of distribution network in real time; a model construction module, configured to, when the network loss data exceeds a preset network loss threshold, use the network active power loss of the distribution network as an objective function, respectively constrain the line flow, distributed power source power, and node voltage in the distribution network, and perform semidefinite programming reconstruction on the resulting constraint conditions, thereby constructing an optimization control model for the distribution network based on the reconstruction results and the objective function; A model decomposition module, configured to obtain coupling branch states between a plurality of edge regions in the distribution network, and decompose the optimization control model into a plurality of distributed optimization control sub-models based on partition coordination based on the coupling branch states; A strategy solving module is used to control the edge agents deployed in each of the edge areas to solve each of the distributed optimization control sub-models using a preset distributed optimization algorithm to obtain a distributed collaborative optimization strategy; The preset distributed optimization algorithm is a synchronous alternating direction multiplier algorithm that introduces the historical iteration results and increments of each edge agent; The control of the edge agents deployed in each edge area uses a preset distributed optimization algorithm to solve each distributed optimization control sub-model to obtain a distributed collaborative optimization strategy, including: Based on each of the distributed optimization control sub-models, control each of the edge agents to construct an augmented Lagrangian objective function and edge constraints for the edge region to which it belongs; Under the edge constraints, control each edge agent to iteratively solve each augmented Lagrangian objective function using the preset distributed optimization algorithm until a preset iterative convergence condition is reached, and output the regional decision variables of each edge region as a distributed collaborative optimization strategy; The controlling each of the edge agents to iteratively solve each of the augmented Lagrangian objective functions using the preset distributed optimization algorithm includes: Initializing the dual variables and penalty parameters of each edge agent, and initializing the coupling branch state and fixed reference value of each edge agent according to the historical iteration results and coupling branch state increments of each edge agent; Solving the augmented Lagrangian objective function of each edge agent using a gradient descent method according to the initialized dual variables, penalty parameters, and fixed reference values ​​to obtain regional decision variables of each edge region; When the regional decision variable does not reach the preset iterative convergence condition, controlling each edge agent to interact with its own regional decision variable with its adjacent edge agent, so that each edge agent updates its own fixed reference value and dual variable according to the regional decision variable of each adjacent edge agent; Updating the penalty parameter by an adaptive penalty parameter adjustment method so that each edge agent solves each augmented Lagrangian objective function based on the updated fixed reference value, dual variable, and penalty parameter to obtain updated regional decision variables; When the updated regional decision variables do not meet the preset iterative convergence condition, repeatedly performing the steps of updating the fixed reference value, the penalty parameter, and the dual variable, and solving the augmented Lagrangian objective function based on the updated regional decision variables until the updated regional decision variables meet the preset iterative convergence condition; The historical iteration results and incremental updates of the edge agents’ own coupling branch states are expressed as follows: Where, 、 are the state increments of the coupling branches in the edge regions a and b respectively; t is the number of iterations; 、 They are the coupling branch states after initialization of edge regions a and b respectively; 、 are the initial values ​​of the coupling branch states of edge regions a and b respectively; m is the number of historical iteration values ​​stored by the edge agent; 、 are the incremental coefficients of edge regions a and b respectively; 、 are the historical increment coefficients corresponding to the edge regions a and b at the i-th historical iteration respectively; The adaptive penalty parameter adjustment method is expressed by the following formula: Where, 、 are the penalty parameters before and after the update respectively; and is an adjustable parameter; is the fixed reference value at the tth iteration.

6. An electronic device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method implements the distributed collaborative optimization control method of the distribution network according to any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored computer program, wherein when the device where the computer-readable storage medium is located executes the computer program, the distributed collaborative optimization control method for the distribution network according to any one of claims 1 to 4 is implemented.

Images