Distributed generator system scheduling optimization method and system based on dual decomposition

Abstract

The invention relates to a distributed generator system scheduling optimization method and system based on dual decomposition, and the method comprises the steps: obtaining the operation data of a power distribution system, including distributed generators and public utilities; the distributed generator is used as a sub-agent, and the public utility is used as a main agent; constructing a scheduling objective function by taking power generation cost minimization as an objective, setting constraint conditions based on operation data, and constructing a scheduling optimization problem; converting the scheduling target problem into a corresponding dual problem by using a Lagrange dual function, and proving that the optimal solutions of the scheduling target problem and the corresponding dual problem are the same based on a Slater condition; the decomposed dual problem comprises a sub-agent optimization problem and a main agent optimization problem; solving a main agent optimization problem by using a subgradient descent method, and carrying out convergence analysis on an optimal solution; iteratively updating the step length and the gradient until a termination criterion is met, and obtaining an optimal scheduling vector; the method has expandability, is suitable for distributed generator deployment, and makes full use of generator capacity.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid dispatching, and mainly to a distributed generator system dispatching optimization method and system based on dual decomposition. Background Art

[0002] The modern power grid is evolving rapidly to meet a variety of end-user and regulatory needs. To meet these needs, it is necessary to utilize multiple sources of power generation, such as solar panels and wind turbines, which provide significant long-term benefits. A technical challenge to the widespread integration of distributed generators is that their generation capacity often fluctuates, making it difficult to apply to scenarios such as power dispatch.

[0003] Existing power dispatch methods can be divided into two categories: offline and online. Offline techniques perform dispatch calculations in advance, while online techniques perform them in shorter time intervals. Centralized offline techniques can conveniently offload the calculation of the optimal dispatch vector to the appropriate resources because these strategies are not subject to strict time constraints. Since offline techniques rely heavily on predictions, an unresolved challenge is that the accuracy of load and generation forecast models tends to decrease as the prediction time increases. In addition, if the power dispatch calculations are too conservative, the excess generation capacity caused by prediction errors will not be utilized.

[0004] For example, the Chinese invention patent with the publication number "CN116307633A" discloses a "Smart Grid Dispatching Method and System Based on Distributed Newton Backward Splitting", which specifically discloses "obtaining the total energy demand of the power grid; obtaining the optimal dispatching result of the unit according to the total energy demand of the power grid and the smart grid dispatching model, wherein the smart grid dispatching model takes the minimum power grid operation cost as the goal and the supply and demand balance constraint as the constraint condition. When the optimal dispatching result of the unit is determined through the smart grid dispatching model, the smart grid dispatching model is converted into a Lagrangian dual problem, and the output of the unit is used as the Lagrangian dual problem. The dual variable of the problem is determined based on the Newton backward splitting algorithm to determine the Lagrange multiplier that maximizes the Lagrange dual value. According to the determined Lagrange multiplier, the dual variable value that minimizes the Lagrange dual value is determined. The dual variable value is the unit optimization scheduling result. However, the Newton backward splitting algorithm used in this method has a relatively complex theoretical background and mathematical derivation. The solution process involves a large amount of calculations and may require a long computing time and high computing resources. In addition, this method may require frequent updates and optimizations of the model to adapt to new demands and resource changes, which increases the maintenance cost and complexity of the model. Summary of the invention

[0005] In order to solve the above problems existing in the prior art, the present application provides a distributed generator system scheduling optimization method and system based on dual decomposition.

[0006] The technical solution of this application is as follows:

[0007] On the one hand, the present invention proposes a distributed generator system scheduling optimization method based on dual decomposition, the method comprising:

[0008] Acquiring operation data of a power distribution system, the power distribution system including a distributed generator and a utility; using the distributed generator as a sub-agent and the utility as a master agent; the operation data including the operation data of the distributed generator and related economic data;

[0009] Constructing a scheduling optimization problem, constructing a scheduling optimization problem, including constructing a corresponding scheduling objective function with the goal of minimizing power generation cost, and constraints set based on operating data; the constraints include power supply capacity, mutual balance of power demand, and scheduling restrictions of distributed generators;

[0010] Based on the Slater condition, it is proved that the optimal solution of the scheduling objective problem and its corresponding dual problem is the same, and the scheduling objective problem is transformed into the corresponding dual problem using the Lagrangian dual function; the dual problem is dually decomposed, and the decomposed dual problem includes a subagent optimization problem and a master agent optimization problem;

[0011] The subagent optimization problem and the master agent optimization problem are solved, wherein the master agent optimization problem is solved by using a subgradient descent method to obtain an optimal solution to the dual problem; and an optimal scheduling strategy for distributed generators is obtained based on the optimal solution.

[0012] Preferably, in the operating data:

[0013] The operating data of the distributed generator includes a dispatch vector, a power generation capacity and the number of dispatch vectors;

[0014] The relevant economic data include the number of loads in the power distribution system, the power requirements of the loads and the power generation cost of the distributed generators.

[0015] Preferably, the corresponding dispatch objective function is constructed with the goal of minimizing the power generation cost, which can be expressed as follows:

[0016]

[0017] x=[x1,x2,...,x n ];

[0018]

[0019] Where P p represents the dispatch objective function; x represents the dispatch vector of distributed generators; x i represents the dispatch vector of the i-th distributed generator; hi represents the generating capacity of the ith distributed generator; n represents the number of dispatch vectors; f0 represents the total generating cost function; subject to represents the constraint condition; P represents the number of loads in the distribution system; W j represents the power demand of the jth load; i represents the index value of the i-th scheduling vector; j represents the index value of the j-th load power demand; C represents the diagonal matrix of power generation cost; C ii represents the power generation cost of the distributed generator in the i-th row and i-th column in the diagonal matrix of power generation cost; min represents the minimum function.

[0020] Preferably, the scheduling objective problem is transformed into a corresponding dual problem using a Lagrangian dual function, which is expressed as follows:

[0021]

[0022]

[0023] λ=[λ1,λ2,...,λ n ];

[0024] Where L represents the Lagrangian dual function; λ represents the first Lagrangian multiplier; λ i represents the i-th first Lagrange multiplier; v represents the second Lagrange multiplier; max represents the maximum value function.

[0025] Preferably, the dual problem is subjected to dual decomposition, and the decomposed dual problem includes a subagent optimization problem and a master agent optimization problem, wherein:

[0026] The subagent optimization problem is specifically to remove the inequality constraint related to the first Lagrange multiplier from the Lagrange dual function, and to treat the inequality constraint related to the first Lagrange multiplier as a local inequality constraint; the subagent optimization problem is expressed as follows:

[0027]

[0028] λ=[λ1,λ2,...,λ n ];

[0029] Where P s represents the subagent optimization function; g i (v) represents the optimal value of the objective function of the subagent optimization function, that is, the optimal local dispatch cost of the i-th generator under a given v;

[0030] The master-agent optimization problem is specifically to simplify the scheduling optimization problem into an unconstrained maximization problem containing only the second Lagrange multiplier, which is expressed as follows:

[0031]

[0032] Where P M represents the master agent optimization function, where P M is a concave function; M(v) represents the objective function value of the dual optimization problem.

[0033] Preferably, the master-agent optimization problem is solved based on the sub-gradient descent method, specifically:

[0034] The sub-gradient descent method specifically iterates the second Lagrange multiplier v, which is expressed as:

[0035] v k+1 =v k +α k+1 q(v k );

[0036] In the formula, v k+1 represents the k+1th iteration of v; q(v k ) represents the sub-gradient of v at the kth iteration; α k+1 represents the step size of the k+1th iteration; k represents the index value of the kth iteration;

[0037] The sub-gradient is specifically the difference between the total dispatch amount and the total power demand, which is expressed as follows:

[0038] M(v k+1 )≤M(v k )+q(v k )(v k+1 -v k );

[0039]

[0040]

[0041]

[0042] In the formula, Indicates g i (v k ) holds true; inf represents the infimum function;

[0043] The master agent optimization function is specifically a concave function. The optimal solution of the dual problem is derived based on the conditions under which the concave function is established, which is expressed as:

[0044]

[0045] M * -M(v k )≤q(vk )(v * -v k );

[0046]

[0047]

[0048] Where R represents the distance range between the preset initial position and the optimal position of v; d k represents the cost function M(v) of M(v) in the kth iteration k ) and the optimal objective function value M * The upper bound of the gap; v * represents the optimal value of v after the optimization of the master-agent problem; M * Indicates that M(v) is at the optimal solution v * The objective function value of ; ||||2 represents the norm; z represents the index value of the zth iteration; Q represents the upper bound of the subgradient.

[0049] Preferably, the step size allocation method includes a constant step size, a decreasing step size and a dynamic step size, wherein:

[0050] The constant step length is specifically defined as a preset constant, which is expressed as follows:

[0051] α k =s;

[0052] In the formula, s represents a preset constant;

[0053] The decreasing step size specifically means that the step size gradually decreases with the number of iterations, which can be expressed as:

[0054]

[0055]

[0056] The dynamic step size is specifically implemented by a path-based target level algorithm to update the step size, which is expressed as follows:

[0057]

[0058] ω k+1 =ω k +α k ·m·Q;

[0059]

[0060]

[0061]

[0062]

[0063] In the formula, ω k represents the oscillation variable; β represents the scaling factor of the preset dynamic step size; ε represents the preset error control parameter; m represents the preset scaling constant; b represents the preset upper limit of the path length;

[0064] The optimal solution is analyzed for convergence, and the step size and gradient are iteratively updated until the termination criterion is met. The optimal scheduling strategy of the distributed generators is obtained according to the optimal solution, where:

[0065] For the constant step size α k =s, we can get the optimal solution convergence;

[0066] Convergence analysis of the decreasing step size can be performed to obtain the optimal solution d k =0 converges, which can be expressed as:

[0067]

[0068] A convergence analysis is performed for the dynamic step size, where:

[0069]

[0070] In the formula, Q is a bounded function, R 2 is a finite value, is convergent, then is a finite value; and Available k is a bounded function, the optimal solution d k =0 converges.

[0071] On the other hand, the present invention also proposes a distributed generator system scheduling optimization system based on dual decomposition, the system comprising a data acquisition module, a scheduling optimization problem construction module, a dual problem generation module, a dual problem solving module and a result output module, wherein:

[0072] The data acquisition module is used to acquire operation data of a power distribution system, the power distribution system includes a distributed generator and a public utility; the distributed generator is used as a sub-agent and the public utility is used as a master agent; the operation data includes the operation data of the distributed generator and related economic data; the operation data is transmitted to a scheduling optimization problem construction module;

[0073] The scheduling optimization problem building module is used to build a scheduling optimization problem, including building a corresponding scheduling objective function with the goal of minimizing power generation cost, and constraints set based on operation data; the constraints include power supply capacity, mutual balance of power demand, and scheduling restrictions of distributed generators;

[0074] The dual problem generation module transforms the scheduling objective problem into a corresponding dual problem using a Lagrange dual function, and proves based on the Slater condition that the optimal solution of the scheduling objective function problem and its corresponding dual problem is the same;

[0075] The dual problem solving module performs dual decomposition on the dual problem, wherein the decomposed dual problem includes a subagent optimization problem and a master agent optimization problem; solves the subagent optimization problem and the master agent optimization problem, wherein the master agent optimization problem is solved by a subgradient descent method to obtain an optimal solution to the dual problem; and obtains an optimal scheduling strategy for distributed generators according to the optimal solution;

[0076] The result output module is used to output the optimal scheduling strategy.

[0077] On the other hand, the present invention further proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the distributed generator system scheduling optimization method based on dual decomposition as described in any embodiment of the present invention is implemented.

[0078] On the other hand, the present invention further proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the distributed generator system scheduling optimization method based on dual decomposition as described in any embodiment of the present invention.

[0079] Compared with the prior art, the present invention has the following beneficial effects:

[0080] 1) The present invention provides a distributed generator system scheduling optimization method and system based on dual decomposition. By constructing a scheduling optimization problem and considering the interaction between distributed generators and utilities, the grid scheduling is made more flexible and can quickly respond to real-time changes and demands in the grid, thereby improving the response speed and accuracy of the grid scheduling.

[0081] 2) The present invention provides a distributed generator system scheduling optimization method and system based on dual decomposition. The dual decomposition method is used to decompose the complex scheduling optimization problem into multiple independent sub-problems, which reduces the complexity of solving the problem and improves the efficiency of calculation. In the dual decomposition process, each sub-agent only needs to disclose the common variables shared with the main agent without disclosing any local information, which protects the integrity of private information and enhances data security.

[0082] 3) The present invention provides a distributed generator system scheduling optimization method and system based on dual decomposition, solves the master-agent optimization problem by the sub-gradient descent method, considers three step size update methods, and performs convergence analysis on the step size to ensure that the algorithm can gradually approach the optimal solution during the iteration process, thereby improving the convergence speed of the algorithm and maintaining good stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 It is a method flow chart of an embodiment of the present invention. DETAILED DESCRIPTION

[0084] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0085] The present invention provides the following technical solution: a distributed generator system scheduling optimization method and system based on dual decomposition.

[0086] Example 1

[0087] See Figure 1 This embodiment provides a distributed generator system scheduling optimization method based on dual decomposition, and the specific steps include:

[0088] S1. Acquire operation data of a power distribution system, the power distribution system including distributed generators and public utilities; use the distributed generators as sub-agents and the public utilities as master agents; the operation data include operation data of the distributed generators and related economic data;

[0089] The operating data of the distributed generator includes a dispatch vector, a power generation capacity and the number of dispatch vectors;

[0090] The relevant economic data include the number of loads in the distribution system, the power demand of the loads and the power generation cost of the distributed generators;

[0091] S2. Constructing a scheduling optimization problem, including constructing a corresponding scheduling objective function with the goal of minimizing power generation cost, and constraints set based on operating data; the constraints include power supply capacity, mutual balance of power demand, and scheduling restrictions of distributed generators;

[0092] S21, the scheduling objective function is expressed as follows:

[0093]

[0094] x=[x1,x2,...,x n ];

[0095]

[0096] Where P p represents the dispatch objective function; x represents the dispatch vector of distributed generators; x i represents the dispatch vector of the i-th distributed generator; h i represents the generating capacity of the ith distributed generator; n represents the number of dispatch vectors; f0 represents the total generating cost function; subject to represents the constraint condition; P represents the number of loads in the distribution system; W j represents the power demand of the jth load; i represents the index value of the i-th scheduling vector; j represents the index value of the j-th load power demand; C represents the diagonal matrix of power generation cost; C ii represents the power generation cost of the distributed generator in the i-th row and i-th column of the diagonal matrix of power generation cost; min represents the minimum function;

[0097] S3. Based on the Slater condition, the scheduling optimization problem is a convex optimization problem, and there is an intersection between the inequality constraint and the equality constraint of the scheduling objective function, which indicates that there is a strong duality between the scheduling optimization problem and its corresponding dual problem, proving that the optimal solution of the scheduling objective problem and its corresponding dual problem is the same;

[0098] S4. Use the Lagrange dual function to transform the scheduling target problem into the corresponding dual problem, which is expressed as follows:

[0099]

[0100] λ=[λ1,λ2,...,λ n ];

[0101] Where L represents the Lagrangian dual function; λ represents the first Lagrangian multiplier; λ i represents the i-th first Lagrange multiplier; v represents the second Lagrange multiplier; max represents the maximum value function;

[0102] S5, performing dual decomposition on the dual problem, wherein the decomposed dual problem includes a subagent optimization problem and a master agent optimization problem;

[0103] S51, the subagent optimization problem is specifically to remove the inequality constraint related to the first Lagrangian multiplier from the Lagrangian dual function, and use the inequality constraint related to the first Lagrangian multiplier as a local inequality constraint; the subagent optimization problem is expressed as follows:

[0104]

[0105] λ=[λ1,λ2,...,λ n ];

[0106] Where P s represents the subagent optimization function; g i (v) represents the optimal value of the objective function of the subagent optimization function, that is, the optimal local dispatch cost of the i-th generator under a given v;

[0107] S52, the master-agent optimization problem is specifically to simplify the scheduling optimization problem into an unconstrained maximization problem containing only the second Lagrange multiplier, which is expressed as:

[0108]

[0109] Where P M represents the master agent optimization function, where P M is a concave function; M(v) represents the objective function value of the dual optimization problem;

[0110] S6. Solve the sub-agent optimization problem and the main-agent optimization problem, wherein the main-agent optimization problem is solved by using a sub-gradient descent method; the sub-gradient descent method is specifically to iterate the second Lagrange multiplier v, which is expressed as:

[0111] v k+1 =v k +α k+1 q(v k );

[0112] In the formula, v k+1 represents the k+1th iteration of v; q(v k ) represents the sub-gradient of v at the kth iteration; α k+1 represents the step size of the k+1th iteration; k represents the index value of the kth iteration;

[0113] The sub-gradient is specifically the difference between the total dispatch amount and the total power demand, which is expressed as follows:

[0114] M(v k+1 )≤M(v k )+q(v k )(v k+1 -v k );

[0115]

[0116]

[0117]

[0118] In the formula, Indicates g i (v k ) holds true; inf represents the infimum function;

[0119] S7, obtaining the optimal solution to the dual problem, wherein the master agent optimization function is specifically a concave function, and the optimal solution to the dual problem is derived based on the condition that the concave function is established, which is expressed by the formula:

[0120]

[0121]

[0122]

[0123] Where R represents the distance range between the preset initial position and the optimal position of v; d k represents the cost function M(v) of M(v) in the kth iteration k ) and the optimal objective function value M * The upper bound of the gap; v * represents the optimal value of v after the optimization of the master-agent problem; M * Indicates that M(v) is at the optimal solution v * The objective function value of is the optimal cost value when the distributed generator scheduling and load demand in the system reach the global optimum by maximizing the objective function M(v) when optimizing the distributed scheduling problem; ||||2 represents the norm; z represents the index value of the zth iteration; Q represents the upper bound of the subgradient;

[0124] S8, using three step size allocation methods of the sub-gradient descent method, specifically constant step size, decreasing step size and dynamic step size;

[0125] S81, the constant step length is specifically defined as a preset constant, which is expressed as:

[0126] α k =s;

[0127] In the formula, s represents a preset constant;

[0128] S82, the decreasing step size is specifically that the step size gradually decreases with the number of iterations, which is expressed as:

[0129]

[0130] S83, the dynamic step size is specifically a path-based target level algorithm to achieve step size update; the path-based target level algorithm is mainly used for dynamic step size calculation, the purpose is to better control the step size adjustment in the sub-gradient method, avoid oscillation and premature convergence, aiming to detect oscillation and gradually approach the target level; firstly, oscillation detection is performed, and the step size is dynamically adjusted by maintaining the oscillation variable, which is expressed as:

[0131]

[0132] ω k+1 =ω k +α k ·m·Q;

[0133] In the formula, ω k represents the oscillation variable; β represents the scaling factor of the preset dynamic step size; ε represents the preset error control parameter; m represents the preset scaling constant;

[0134] Secondly, the target level is updated to ensure that the objective function value converges to the optimal value, which can be expressed as:

[0135]

[0136] If the current objective function value exceeds the target level, it can be expressed as:

[0137]

[0138] If the path length exceeds the upper limit, it can be expressed as:

[0139]

[0140] In other cases, the formula is:

[0141]

[0142] Where b represents the upper limit of the path length, the path refers to the trajectory taken by the path-based target level algorithm when adjusting the target value in multiple iterations, and the path length represents the step length from the initial value α 0 To the current iteration α k the amount of cumulative change experienced;

[0143] S9, performing convergence analysis on the optimal solution, iteratively updating the step size and gradient until the termination criterion is met, and obtaining the optimal scheduling strategy of the distributed generators according to the optimal solution, wherein:

[0144] For the constant step size α k =s, we can get the optimal solution convergence;

[0145] Convergence analysis of the decreasing step size can be performed to obtain the optimal solution d k =0 converges, which can be expressed as:

[0146]

[0147] A convergence analysis is performed for the dynamic step size, where:

[0148]

[0149] In the formula, Q is a bounded function, R 2 is a finite value, is convergent, then is a finite value; and Available k is a bounded function, the optimal solution d k = 0 convergence;

[0150] S10, iteratively updating the step size and gradient until a termination criterion is met to obtain an optimal solution; the termination criterion is specifically that when the sub-gradient is 0, the total dispatch amount of the distributed generators is equal to the total demand of the distribution system; and obtaining an optimal dispatch strategy for the distributed generators according to the optimal solution;

[0151] Since the subagent optimization problem is specifically a function of the scheduling vector x and the second Lagrangian multiplier v: and the master agent optimization problem is specifically a function containing only the second Lagrangian multiplier v:

[0152] By finding the optimal solution to the master-agent optimization problem, the optimal second Lagrange multiplier v can be obtained, so the scheduling vector x in the sub-agent optimization problem can be solved and the optimal scheduling strategy of the distributed generators can be obtained.

[0153] Example 2

[0154] This embodiment provides a distributed generator system scheduling optimization system based on dual decomposition, the system includes a data acquisition module, a scheduling optimization problem construction module, a dual problem generation module, a dual problem solution module and a result output module, wherein:

[0155] The data acquisition module is used to acquire operation data of a power distribution system, the power distribution system includes a distributed generator and a public utility; the distributed generator is used as a sub-agent and the public utility is used as a master agent; the operation data includes the operation data of the distributed generator and related economic data; the operation data is transmitted to a scheduling optimization problem construction module;

[0156] The scheduling optimization problem building module is used to build a scheduling optimization problem, including building a corresponding scheduling objective function with the goal of minimizing power generation cost, and constraints set based on operation data; the constraints include power supply capacity, mutual balance of power demand, and scheduling restrictions of distributed generators;

[0157] The dual problem generation module transforms the scheduling objective problem into a corresponding dual problem using a Lagrange dual function, and proves based on the Slater condition that the optimal solution of the scheduling objective function problem and its corresponding dual problem is the same;

[0158] The dual problem solving module performs dual decomposition on the dual problem, wherein the decomposed dual problem includes a subagent optimization problem and a master agent optimization problem; solves the subagent optimization problem and the master agent optimization problem, wherein the master agent optimization problem is solved by a subgradient descent method to obtain an optimal solution to the dual problem; performs convergence analysis on the optimal solution by using three step size allocation methods of the subgradient descent method; iteratively updates the step size and gradient until a termination criterion is met, and obtains an optimal scheduling strategy for distributed generators according to the optimal solution;

[0159] The result output module is used to output the optimal scheduling strategy.

[0160] Example 3

[0161] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the distributed generator system scheduling optimization method based on dual decomposition as described in any embodiment of the present invention is implemented.

[0162] Example 4

[0163] This embodiment provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the distributed generator system scheduling optimization method based on dual decomposition as described in any embodiment of the present invention is implemented.

[0164] It is worth noting that the system, electronic device and computer-readable storage medium described in the present invention are based on the same principles as the method described in Example 1, and will not be described in detail here.

[0165] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A distributed generator system scheduling optimization method based on dual decomposition, characterized in that: The method comprises: Acquiring operation data of a power distribution system, the power distribution system including a distributed generator and a utility; using the distributed generator as a sub-agent and the utility as a master agent; the operation data including the operation data of the distributed generator and related economic data; Constructing a scheduling optimization problem, including constructing a corresponding scheduling objective function with the goal of minimizing power generation cost, and setting constraints based on operating data; the constraints include power supply capacity, mutual balance of power demand, and scheduling restrictions of distributed generators; Based on the Slater condition, it is proved that the optimal solution of the scheduling objective problem and its corresponding dual problem is the same, and the scheduling objective problem is transformed into the corresponding dual problem using the Lagrangian dual function; the dual problem is dually decomposed, and the decomposed dual problem includes a subagent optimization problem and a master agent optimization problem; The subagent optimization problem and the master agent optimization problem are solved, wherein the master agent optimization problem is solved by using a subgradient descent method to obtain an optimal solution to the dual problem; and an optimal scheduling strategy for distributed generators is obtained based on the optimal solution.

2. The distributed generator system scheduling optimization method based on dual decomposition according to claim 1 is characterized in that: In the operating data: The operating data of the distributed generator includes a dispatch vector, a power generation capacity and the number of dispatch vectors; The relevant economic data include the number of loads in the power distribution system, the power requirements of the loads and the power generation cost of the distributed generators.

3. The distributed generator system scheduling optimization method based on dual decomposition according to claim 2 is characterized in that: The corresponding dispatch objective function is constructed with the goal of minimizing the power generation cost, which can be expressed as follows: 0≤x i ≤h i ,i=1,2,...,n x=[x1,x2,...,x n ]; Where P p represents the dispatch objective function; x represents the dispatch vector of distributed generators; x i represents the dispatch vector of the i-th distributed generator; h i represents the generating capacity of the ith distributed generator; n represents the number of dispatch vectors; f0 represents the total generating cost function; subject to represents the constraint condition; P represents the number of loads in the distribution system; W j represents the power demand of the jth load; i represents the index value of the i-th scheduling vector; j represents the index value of the j-th load power demand; C represents the diagonal matrix of power generation cost; C ii represents the power generation cost of the distributed generator in the i-th row and i-th column in the diagonal matrix of power generation cost; min represents the minimum function.

4. The distributed generator system scheduling optimization method based on dual decomposition according to claim 3 is characterized in that: The Lagrange dual function is used to transform the scheduling objective problem into the corresponding dual problem, which is expressed as follows: λ=[λ1,λ2,...,λ n ]; Where L represents the Lagrangian dual function; λ represents the first Lagrangian multiplier; λ i represents the i-th first Lagrange multiplier; v represents the second Lagrange multiplier; max represents the maximum value function.

5. The distributed generator system scheduling optimization method based on dual decomposition according to claim 4 is characterized in that: The dual problem is dually decomposed, and the decomposed dual problem includes a subagent optimization problem and a master agent optimization problem, wherein: The subagent optimization problem is specifically to remove the inequality constraint related to the first Lagrange multiplier from the Lagrange dual function, and to treat the inequality constraint related to the first Lagrange multiplier as a local inequality constraint; the subagent optimization problem is expressed as follows: subject to 0≤x i ≤h i λ=[λ1,λ2,...,λ n ]; Where P s represents the subagent optimization function; g i (v) represents the optimal value of the objective function of the subagent optimization function, that is, the optimal local dispatch cost of the i-th generator under a given v; The master-agent optimization problem is specifically to simplify the scheduling optimization problem into an unconstrained maximization problem containing only the second Lagrange multiplier, which is expressed as follows: Where P M represents the master agent optimization function, where P M is a concave function; M(v) represents the objective function value of the dual optimization problem.

6. The distributed generator system scheduling optimization method based on dual decomposition according to claim 5 is characterized in that: The master-agent optimization problem is solved based on the sub-gradient descent method, specifically: The sub-gradient descent method specifically iterates the second Lagrange multiplier v, which is expressed as: v k+1 =v k +α k+1 q(v k ); In the formula, v k+1 represents the k+1th iteration of v; q(v k ) represents the sub-gradient of v at the kth iteration; α k+1 represents the step size of the k+1th iteration; k represents the index value of the kth iteration; The sub-gradient is specifically the difference between the total dispatch amount and the total power demand, which is expressed as follows: M(v k+1 )≤M(v k )+q(v k )(v k+1 -v k ); In the formula, Indicates g i (v k ) holds true; inf represents the infimum function; The master agent optimization function is specifically a concave function. The optimal solution of the dual problem is derived based on the conditions under which the concave function is established, which is expressed as: M * -M(in k )≤q(v k )(in * -v k ); Where R represents the distance range between the preset initial position and the optimal position of v; d k represents the cost function M(v) of M(v) in the kth iteration k ) and the optimal objective function value M * The upper bound of the gap; v * represents the optimal value of v after the optimization of the master-agent problem; M * Indicates that M(v) is at the optimal solution v * The objective function value of ; || ||2 represents the norm; z represents the index value of the zth iteration; Q represents the upper bound of the subgradient.

7. The distributed generator system scheduling optimization method based on dual decomposition according to claim 6 is characterized in that: The step size allocation method includes a constant step size, a decreasing step size and a dynamic step size, wherein: The constant step length is specifically defined as a preset constant, which is expressed as follows: α k ＝s； In the formula, s represents a preset constant; The decreasing step size specifically means that the step size gradually decreases with the number of iterations, which can be expressed as: The dynamic step size is specifically implemented by a path-based target level algorithm to update the step size, which is expressed as follows: oh k+1 =ω k +a k ·m·Q; In the formula, ω k represents the oscillation variable; β represents the scaling factor of the preset dynamic step size; ε represents the preset error control parameter; m represents the preset scaling constant; b represents the preset upper limit of the path length; The optimal solution is analyzed for convergence, and the step size and gradient are iteratively updated until the termination criterion is met. The optimal scheduling strategy of the distributed generators is obtained according to the optimal solution, where: For the constant step size α k =s, we can get the optimal solution convergence; Convergence analysis of the decreasing step size can be performed to obtain the optimal solution d k =0 converges, expressed as: A convergence analysis is performed for the dynamic step size, where: In the formula, Q is a bounded function, R 2 is a finite value, is convergent, then is a finite value; and Available k is a bounded function, the optimal solution d k =0 converges.

8. Distributed generator system scheduling optimization system based on dual decomposition, characterized in that: The system includes a data acquisition module, a scheduling optimization problem construction module, a dual problem generation module, a dual problem solution module and a result output module, wherein: The data acquisition module is used to acquire operation data of a power distribution system, the power distribution system includes a distributed generator and a public utility; the distributed generator is used as a sub-agent and the public utility is used as a master agent; the operation data includes the operation data of the distributed generator and related economic data; the operation data is transmitted to a scheduling optimization problem construction module; The scheduling optimization problem building module is used to build a scheduling optimization problem, including building a corresponding scheduling objective function with the goal of minimizing power generation cost, and constraints set based on operation data; the constraints include power supply capacity, mutual balance of power demand, and scheduling restrictions of distributed generators; The dual problem generation module transforms the scheduling objective problem into a corresponding dual problem using a Lagrange dual function, and proves based on the Slater condition that the optimal solution of the scheduling objective function problem and its corresponding dual problem is the same; The dual problem solving module performs dual decomposition on the dual problem, wherein the decomposed dual problem includes a subagent optimization problem and a master agent optimization problem; solves the subagent optimization problem and the master agent optimization problem, wherein the master agent optimization problem is solved by a subgradient descent method to obtain an optimal solution to the dual problem; and obtains an optimal scheduling strategy for distributed generators according to the optimal solution; The result output module is used to output the optimal scheduling strategy.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the distributed generator system scheduling optimization method based on dual decomposition as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the distributed generator system scheduling optimization method based on dual decomposition as described in any one of claims 1 to 7 is implemented.

Images