Search

CN-122026478-A - Virtual power plant spare capacity optimal configuration management method and system

CN122026478ACN 122026478 ACN122026478 ACN 122026478ACN-122026478-A

Abstract

The invention provides a virtual power plant spare capacity optimizing configuration management method and system, which relate to the technical field of power system management and comprise the steps of obtaining distributed energy unit parameters and power grid demand information, clustering and grouping are carried out based on the response time delay-energy state, a double-layer optimization framework is constructed, nash equilibrium solutions are solved by adopting a multi-main-body non-cooperative game model in the cluster, and a final spare capacity configuration scheme is obtained through upper and lower layer configuration quantity deviation convergence. The scheme realizes the effective coordination of heterogeneous resources in the virtual power plant, improves the spare capacity configuration efficiency and reduces the system operation cost.

Inventors

  • PAN YINGCHAO
  • DUAN XIAOHAN

Assignees

  • 北京如实智慧电力科技有限公司

Dates

Publication Date
20260512
Application Date
20260203

Claims (10)

  1. 1. The virtual power plant spare capacity optimizing configuration management method is characterized by comprising the following steps: Acquiring adjustable capacity, response time delay and energy state of each distributed energy unit in the virtual power plant and spare capacity demand information of a power grid side; Based on the response time delay and the energy state, clustering and grouping the distributed energy units in a response time delay-energy state two-dimensional space to obtain a plurality of state clustering clusters, calculating the mass center coordinates of the state clustering clusters and the quantity of resources in the clusters, and constructing a state clustering cluster mapping table; Constructing a double-layer optimization framework based on the state cluster mapping table, wherein the upper-layer optimization aims at maximizing the operation income of the virtual power plant, optimizing the allocation target quantity of the spare capacity of each state cluster, and the lower-layer optimization aims at minimizing the intra-cluster response cost of each state cluster and optimizing the actual allocation quantity; Aiming at each state cluster, constructing a multi-main-body non-cooperative game model in the cluster based on the distribution target quantity and the centroid coordinates, obtaining Nash equilibrium solutions by solving optimal response functions of game main bodies, and calculating initial configuration quantities of distributed energy units according to the equilibrium solutions; Substituting the initial configuration quantity into the lower-layer optimization, extracting a KKT multiplier as a feedback signal to adjust the distribution target quantity of the upper-layer optimization, and re-executing intra-cluster game decomposition until the deviation of the upper-layer configuration quantity and the lower-layer configuration quantity converges to obtain a final spare capacity configuration scheme; And issuing a spare capacity configuration instruction to each distributed energy unit based on the final spare capacity configuration scheme.
  2. 2. The method of claim 1, wherein clustering each distributed energy unit in a response delay-energy state two-dimensional space based on the response delay and energy state to obtain a plurality of state clusters, and calculating centroid coordinates of each state cluster and the number of resources in the cluster to construct a state cluster map table, comprising: Determining the spatial position point of each distributed energy unit in a response time delay-energy state two-dimensional space by taking the response time delay of each distributed energy unit as a first-dimensional coordinate and the energy state as a second-dimensional coordinate; Performing iterative grouping on the spatial position points by adopting a density and distance-based double-constraint clustering algorithm, and calculating the weighted distance from each spatial position point to each current clustering centroid in each iteration, wherein the weighted distance comprises a time similarity weight and a capacity similarity weight; distributing each distributed energy unit to a corresponding state cluster according to the weighted distance, recalculating the centroid coordinates of each state cluster, and stopping iteration when the centroid coordinate variation of each state cluster in continuous iteration is smaller than a convergence threshold value to obtain a plurality of final state clusters; for each state clustering cluster, calculating the coordinate mean value of the distributed energy units in the cluster, taking the coordinate mean value as a centroid coordinate, counting the number of the units in the cluster, establishing the association relation among the centroid coordinate, the number of the units in the cluster and each distributed energy unit identifier in the cluster, and constructing a state clustering cluster mapping table.
  3. 3. The method of claim 1, wherein constructing a double-layer optimization framework based on the state cluster mapping table, the upper-layer optimization targeting the maximization of virtual power plant operation yield, optimizing the allocation target amount of spare capacity of each state cluster, the lower-layer optimization targeting the minimization of intra-cluster response cost of each state cluster, optimizing the actual configuration amount, comprises: Based on the state cluster mapping table, constructing a response time delay-energy state coupling matrix, carrying out Cartesian product operation on the barycenter coordinates to obtain a coupling feature vector, and calculating the inner product of the coupling feature vector and the spare capacity requirement to obtain a requirement matching degree; Taking the allocation target quantity of the spare capacity of each state cluster as a decision variable, constructing an upper optimization model, wherein the upper optimization target function comprises a ratio deviation metric item of the allocation target quantity and the required matching degree, and a boundary transition constraint item of the allocation target quantity between adjacent clusters, and the constraint condition is that the allocation total quantity meets the spare requirement and the allocation quantity of each cluster is non-negative; For each state cluster, constructing a cluster content distribution reference function based on the distribution target quantity and the mass center coordinates, and mapping the response time delay characteristic and the energy state of each distributed energy unit into a capacity distribution reference value; And constructing a lower-layer optimization model, taking the actual configuration quantity of each distributed energy unit in the cluster as a decision variable, wherein a lower-layer optimization objective function comprises a deviation term of the actual configuration quantity and the capacity allocation reference value and a coupling term of response cost, and ensuring that the total configuration quantity in the cluster is equal to the allocation target quantity and does not exceed the adjustable capacity.
  4. 4. The method of claim 1, wherein for each state cluster, constructing an intra-cluster multi-subject non-cooperative game model based on the allocation target amount and the centroid coordinates, obtaining a nash equilibrium solution by solving an optimal response function of each game subject, and calculating an initial configuration amount of each distributed energy unit according to the equilibrium solution, comprising: for each state cluster, taking each distributed energy unit in the cluster as a game main body, calculating a time delay competition factor and a capacity competition factor according to the distribution target quantity and the mass center coordinates, and carrying out tensor product operation to obtain a two-dimensional competition intensity matrix of each state cluster; establishing a profit function for each game main body, wherein the profit function comprises a product term of configuration quantity and unit capacity profit, and a bilinear competition loss term based on a two-dimensional competition intensity matrix, and performing bias guide on the profit function relative to the configuration quantity of each game main body, and enabling the profit function to be equal to zero to obtain an optimal response function of each game main body relative to the configuration quantity of other game main bodies; Combining the optimal response functions of all game main bodies to form a nonlinear equation set, introducing the allocation target quantity as a total constraint condition, and solving the nonlinear equation set through iteration to obtain a Nash equilibrium solution meeting the requirement that the optimal response functions of all game main bodies are simultaneously established; And extracting each component in the Nash equilibrium solution, and taking each component as the initial configuration quantity of each distributed energy unit in the state cluster.
  5. 5. The method of claim 4 wherein combining the optimal response functions of each gaming entity to form a system of nonlinear equations and introducing the allocation target as a total constraint, and solving the system of nonlinear equations iteratively to obtain a nash equalization solution that satisfies simultaneous establishment of all gaming entity optimal response functions comprises: combining the optimal response functions of all game main bodies to form a nonlinear equation set, extracting a configuration quantity coupling coefficient matrix, and performing precondition transformation based on the spectrum radius and condition number of the two-dimensional competition strength matrix to obtain a positively-defined symmetrical preprocessing matrix; constructing an augmented Lagrangian function by taking the allocation target quantity as constraint, wherein the augmented Lagrangian function comprises a nonlinear equation set residual term, a preprocessing matrix weighted quadratic penalty term and a linear coupling term of Lagrangian multiplier and constraint deviation; decomposing the augmented Lagrangian function into a configuration quantum problem and a multiplier updating sub-problem by an alternate direction multiplier method, introducing an inverse matrix of a pretreatment matrix as a descending direction correction operator, and setting uniform allocation quantity as an iteration initial value; And sequentially solving the configuration quantum problem in the iteration process to obtain an updated value, updating the Lagrange multiplier based on the deviation between the updated value and the target quantity, calculating the residual norm of the nonlinear equation set, and outputting the current configuration quantity as a Nash equilibrium solution when the residual norm is smaller than a residual threshold and the configuration quantity updated value meets the total constraint.
  6. 6. The method of claim 1, wherein substituting the initial configuration quantity into the lower layer optimization, extracting a KKT multiplier as a feedback signal to adjust an allocation target quantity of the upper layer optimization, and re-executing intra-cluster game decomposition until deviation of the upper and lower layer configuration quantities converges, to obtain a final spare capacity configuration scheme, comprising: Substituting the initial configuration quantity into a lower-layer optimization objective function, constructing a KKT optimality condition equation set by combining constraint conditions, solving to obtain a KKT multiplier corresponding to the distribution target quantity constraint, and constructing a sensitivity vector representing the marginal response intensity of the actual configuration quantity based on the sign and the numerical value of the KKT multiplier; performing inner product operation on the sensitivity vector and the gradient vector of the upper-layer optimization objective function to obtain an adjustment direction factor, determining a self-adaptive step length based on a module value of the adjustment direction factor, correcting an upper-layer optimization allocation objective quantity to obtain a corrected allocation objective quantity, substituting the corrected allocation objective quantity into a multi-main-body non-cooperative game model in a cluster, performing game decomposition to obtain an updated initial configuration quantity, and substituting the updated initial configuration quantity into a lower-layer optimization calculation new actual configuration quantity again; Calculating a deviation vector norm between the updated initial configuration quantity and the new actual configuration quantity, outputting the new actual configuration quantity as a final scheme when the deviation is smaller than a preset threshold, otherwise, repeatedly executing iterative processes of KKT multiplier extraction, sensitivity calculation and target quantity correction based on the new actual configuration quantity until convergence.
  7. 7. The method of claim 6 wherein substituting the initial configuration quantity into a lower layer optimization objective function, constructing a system of KKT optimality conditional equations in combination with constraints, solving to obtain a KKT multiplier corresponding to an allocation objective quantity constraint, and constructing a sensitivity vector representing a marginal response strength of an actual configuration quantity based on a sign and a value of the KKT multiplier, comprising: substituting the initial configuration quantity into a lower-layer optimization objective function, constructing an augmentation objective functional through a penalty term to soften inequality constraint, solving first-order essential conditions and combining complementary relaxation conditions to construct a KKT optimality equation set; Performing linear decomposition on the KKT optimality condition equation set, extracting a sensitive submatrix corresponding to the allocation target quantity constraint, decomposing the sensitive submatrix into a product form of an orthogonal matrix and an upper triangular matrix, and obtaining KKT multipliers of each cluster by using back substitution solution; Carrying out symbol discrimination on the KKT multiplier to obtain an indication vector, carrying out logarithmic transformation on the absolute value of the indication vector to obtain an amplitude vector, carrying out element-by-element multiplication to obtain a logarithmic domain sensitivity vector, carrying out nonlinear scaling on the logarithmic domain sensitivity vector, and carrying out exponential inverse transformation mapping to obtain a sensitivity vector representing the marginal response intensity of the actual configuration quantity.
  8. 8. A virtual power plant backup capacity optimization configuration management system for implementing the method of any one of claims 1-7, comprising: the resource information acquisition unit is used for acquiring the adjustable capacity, response time delay and energy state of each distributed energy unit in the virtual power plant and the spare capacity demand information of the power grid side; the state space clustering unit is used for clustering and grouping the distributed energy units in a response time delay-energy state two-dimensional space based on the response time delay and the energy state to obtain a plurality of state clustering clusters, calculating the mass center coordinates of the state clustering clusters and the quantity of resources in the clusters, and constructing a state clustering cluster mapping table; The double-layer optimization decision unit is used for constructing a double-layer optimization framework based on the state cluster mapping table, wherein the upper-layer optimization aims at maximizing the operation income of the virtual power plant, optimizing the allocation target quantity of the spare capacity of each state cluster, and the lower-layer optimization aims at minimizing the intra-cluster response cost of each state cluster and optimizing the actual allocation quantity; The in-cluster game balancing unit is used for constructing a multi-main-body non-cooperative game model in the cluster based on the distribution target quantity and the mass center coordinates aiming at each state cluster, obtaining Nash balancing solutions by solving optimal response functions of all game main bodies, and calculating initial configuration quantities of all distributed energy units according to the balancing solutions; the iteration convergence coordination unit is used for substituting the initial configuration quantity into the lower-layer optimization, extracting a KKT multiplier as a feedback signal to adjust the allocation target quantity of the upper-layer optimization, and re-executing intra-cluster game decomposition until the deviation convergence of the upper-layer configuration quantity and the lower-layer configuration quantity, so as to obtain a final spare capacity configuration scheme; And the instruction distribution execution unit is used for issuing spare capacity configuration instructions to each distributed energy unit based on the final spare capacity configuration scheme.
  9. 9. An electronic device, comprising: A processor; A memory for storing processor-executable instructions; Wherein the processor is configured to invoke the instructions stored in the memory to perform the method of any of claims 1 to 7.
  10. 10. A computer readable storage medium having stored thereon computer program instructions, which when executed by a processor, implement the method of any of claims 1 to 7.

Description

Virtual power plant spare capacity optimal configuration management method and system Technical Field The invention relates to the technical field of power system management, in particular to a virtual power plant spare capacity optimal configuration management method and system. Background With the large-scale access of renewable energy sources and the wide application of distributed energy sources, the operation characteristics of power systems have changed deeply. Virtual power plants are taken as a novel energy aggregation form, and a large number of scattered distributed energy units are integrated and scheduled through information and communication technology, so that the virtual power plants become an important means for promoting renewable energy consumption and improving flexibility of an electric power system. Particularly, in the background of increasing demand for backup capacity on the grid side, the virtual power plant can effectively support safe and stable operation of the power system by aggregating different types of distributed energy units to provide backup service. In the conventional virtual power plant spare capacity configuration method, a centralized optimization method is generally adopted, and the spare capacity configuration of each unit is determined by solving the overall optimization problem according to the technical characteristics and cost parameters of each distributed energy unit. With the expansion of the scale of the virtual power plant and the diversification of the distributed energy source types, the resources show great heterogeneity in response characteristics, energy states and the like, so that the optimal configuration of the spare capacity of the virtual power plant faces new challenges. Disclosure of Invention The embodiment of the invention provides a virtual power plant spare capacity optimal configuration management method and system, which can solve the problems in the prior art. In a first aspect of the embodiment of the present invention, a method for optimizing configuration management of a backup capacity of a virtual power plant is provided, including: Acquiring adjustable capacity, response time delay and energy state of each distributed energy unit in the virtual power plant and spare capacity demand information of a power grid side; Based on the response time delay and the energy state, clustering and grouping the distributed energy units in a response time delay-energy state two-dimensional space to obtain a plurality of state clustering clusters, calculating the mass center coordinates of the state clustering clusters and the quantity of resources in the clusters, and constructing a state clustering cluster mapping table; Constructing a double-layer optimization framework based on the state cluster mapping table, wherein the upper-layer optimization aims at maximizing the operation income of the virtual power plant, optimizing the allocation target quantity of the spare capacity of each state cluster, and the lower-layer optimization aims at minimizing the intra-cluster response cost of each state cluster and optimizing the actual allocation quantity; Aiming at each state cluster, constructing a multi-main-body non-cooperative game model in the cluster based on the distribution target quantity and the centroid coordinates, obtaining Nash equilibrium solutions by solving optimal response functions of game main bodies, and calculating initial configuration quantities of distributed energy units according to the equilibrium solutions; Substituting the initial configuration quantity into the lower-layer optimization, extracting a KKT multiplier as a feedback signal to adjust the distribution target quantity of the upper-layer optimization, and re-executing intra-cluster game decomposition until the deviation of the upper-layer configuration quantity and the lower-layer configuration quantity converges to obtain a final spare capacity configuration scheme; And issuing a spare capacity configuration instruction to each distributed energy unit based on the final spare capacity configuration scheme. Based on the response time delay and the energy state, clustering and grouping the distributed energy units in a response time delay-energy state two-dimensional space to obtain a plurality of state clustering clusters, calculating the mass center coordinates of the state clustering clusters and the quantity of resources in the clusters, and constructing a state clustering cluster mapping table, wherein the method comprises the following steps: Determining the spatial position point of each distributed energy unit in a response time delay-energy state two-dimensional space by taking the response time delay of each distributed energy unit as a first-dimensional coordinate and the energy state as a second-dimensional coordinate; Performing iterative grouping on the spatial position points by adopting a density and distance-based double-constraint clustering alg