A virtual power plant multi-agent collaborative optimization scheduling method and system

Abstract

The application discloses a kind of virtual power plant multi-agent collaborative optimization scheduling method and system, belong to virtual power plant scheduling optimization technical field.The technical problem of this application is that the heterogeneity of multi-agent is not accurately described in the scheduling of virtual power plant, the adjustment target is single, the uncertainty adaptation is insufficient and the solution efficiency is low.By obtaining the heterogeneity physical characteristic data of multi-agent and processing to obtain the characteristic parameters of each agent, a comprehensive adjustment potential index is obtained based on the characteristic parameters through a multi-dimensional adjustment potential evaluation model.A multi-agent collaborative scheduling strategy is obtained by processing the potential index through a collaborative interaction adjustment model.The optimal operation plan of virtual power plant is obtained by solving through an improved whale-differential evolution fusion algorithm.Finally, the adjustment potential of each agent is accurately excavated, the operation economy, flexibility and environmental protection are considered, and the solution efficiency of complex model is significantly improved.

Description

Technical Field

[0001] This invention relates to the field of virtual power plant scheduling optimization technology, and in particular to a multi-entity collaborative optimization scheduling method and system for virtual power plants. Background Technology

[0002] The statements in this section are merely to provide background information related to the present invention and do not necessarily constitute prior art.

[0003] With the large-scale integration of distributed photovoltaic (PV) power, user-side energy storage, adjustable loads, and electric vehicles into the distribution network, virtual power plants have become a key carrier for aggregating these dispersed resources to participate in grid interaction. However, in actual operation, the characteristics of various resources vary greatly: PV output fluctuates drastically with the weather, frequent charging and discharging of energy storage batteries accelerates their lifespan degradation, the air conditioning load adjustment of commercial buildings is constrained by user comfort, and the connection time and power demand of electric vehicles are highly random.

[0004] Traditional regulation methods often employ a uniform model for these resources, allocating regulation tasks solely based on rated power or a fixed proportion. This leads to frequent problems such as neglecting the true lifespan loss of energy storage, electric vehicles being unable to respond to dispatch due to travel demand, and user dissatisfaction arising from load regulation. Furthermore, existing strategies largely focus on reducing electricity purchase costs, neglecting the potential benefits of virtual power plants providing peak shaving and valley filling during grid peak periods and reducing energy curtailment during periods of high renewable energy generation. Moreover, traditional optimization algorithms are slow to compute and prone to getting trapped in local optima when faced with real-time weather and load fluctuations, making it difficult to quickly generate reliable regulation commands. Therefore, existing technologies do not fully consider the heterogeneous characteristics of each entity, cannot accurately tap its regulation potential, and struggle to balance economy and flexibility while achieving rapid solutions. This could potentially lead to ineffective actual regulation commands, resource waste, or insufficient power supply. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a multi-entity collaborative optimization scheduling method and system for virtual power plants. By accurately characterizing the capability boundaries of each entity, combining a multi-dimensional evaluation model to explore adjustment potential, and introducing an improved whale-differential evolution fusion algorithm to efficiently solve the multi-objective collaborative model, this invention achieves precise collaborative scheduling of distributed power sources, energy storage, loads, and electric vehicles within the virtual power plant, thereby improving operational economy, flexibility, and solution efficiency.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution: In a first aspect, the present invention provides a multi-agent collaborative optimization scheduling method for a virtual power plant, comprising: Acquire heterogeneous physical characteristic data of multiple entities within a virtual power plant; The heterogeneous physical characteristic data are processed to obtain the characteristic parameters of each subject; based on the characteristic parameters, a comprehensive adjustment potential index is obtained through a multi-dimensional adjustment potential assessment model; the comprehensive adjustment potential index is processed through a collaborative interaction adjustment model to obtain a multi-subject collaborative scheduling strategy. The optimal operation plan for the virtual power plant is obtained by solving the multi-agent collaborative scheduling strategy using an improved whale-differential evolution fusion algorithm.

[0007] Preferably, heterogeneous physical characteristic data are processed by establishing a multi-entity heterogeneous characteristic quantification model of a virtual power plant; the multi-entity includes at least distributed power sources, energy storage devices, adjustable loads, and electric vehicles.

[0008] Preferably, the virtual power plant multi-agent heterogeneity characteristic quantification model includes at least: The model includes a distributed power output model incorporating a fluctuation coefficient; a lifespan loss model for energy storage devices that includes charge / discharge efficiency correction and deep charge / discharge loss coefficient; an adjustable load regulation capability model that includes response delay coefficient and user comfort constraint coefficient; and a dispatchable capacity model for electric vehicles that includes grid rate and target state of charge correction.

[0009] Preferably, the multi-dimensional regulation potential assessment model constructs assessment indicators from four dimensions: regulation capacity, response speed, regulation economy, and environmental friendliness. The weights of each indicator were determined using the entropy weight-analytic hierarchy process to calculate the comprehensive adjustment potential index.

[0010] Preferably, the collaborative and interactive regulation model includes a multi-objective function that minimizes total operating cost, maximizes flexibility value, and maximizes environmental benefits, as well as main body regulation power constraints, regulation task matching constraints, main body operation constraints, and grid interaction constraints; Among them, the value of flexibility includes peak shaving and valley filling, backup support, and reduced energy curtailment.

[0011] Preferably, the improved whale-differential evolutionary fusion algorithm includes: The regulation power of each entity and the power grid interaction power of the virtual power plant are encoded into the position vector of the individual whale. Initialize the population position using a chaotic logistic mapping; Introducing adaptive weights and the Levy flight strategy to update individual positions; Perform differential evolution mutation and crossover operations on the population updated by the whale algorithm; Iterate until the termination condition is met, and output the globally optimal position vector as the optimal operation plan for the virtual power plant.

[0012] Preferably, when solving the multi-agent cooperative scheduling strategy using the improved whale-differential evolution fusion algorithm, a repair-penalty hybrid strategy is adopted to handle constraint violations: The power regulation limit exceedance constraint is repaired using a truncation function; the response speed constraint and the power interaction constraint with the power grid are handled using a penalty function.

[0013] In a second aspect, the present invention provides a virtual power plant multi-entity collaborative optimization scheduling system, comprising: Data acquisition module: used to acquire heterogeneous physical characteristic data of multiple entities within the virtual power plant; The characteristic quantization module is used to process the heterogeneous physical characteristic data to obtain the characteristic parameters of each subject; The potential assessment module is used to obtain a comprehensive adjustment potential index based on the aforementioned characteristic parameters through a multi-dimensional adjustment potential assessment model. The collaborative scheduling module is used to process the comprehensive adjustment potential index through a collaborative interactive adjustment model to obtain a multi-entity collaborative scheduling strategy. The optimization solution module is used to solve the multi-agent collaborative scheduling strategy by improving the whale-differential evolution fusion algorithm to obtain the optimal operation plan of the virtual power plant.

[0014] In a third aspect, the present invention also provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method as described in the first aspect.

[0015] In a fourth aspect, the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method as described in the first aspect.

[0016] Compared with the prior art, the beneficial effects of the present invention are: This invention acquires and processes heterogeneous physical characteristic data to accurately characterize the true capability boundaries of each entity in terms of output fluctuations, lifetime loss, and response delays, solving the potential mismatch problem caused by traditional homogenization assumptions. Based on characteristic parameters, it quantifies the regulation potential through a multi-dimensional evaluation model, ensuring optimal suitability between regulation tasks and entities. Then, the potential index is input into a collaborative interaction regulation model to generate a collaborative scheduling strategy within a multi-objective framework that considers operating costs, flexibility value, and environmental benefits, overcoming the limitations of a single cost objective. Finally, an improved whale-differential evolutionary fusion algorithm is used for efficient solution, ensuring the globally optimal operating plan is obtained within the real-time scheduling timescale. This achieves accurate mining of regulation potential, a balance between the economy and flexibility of scheduling schemes, and rapid solution of complex models, significantly improving the operational efficiency of virtual power plants and the grid response capability. Attached Figure Description

[0017] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0018] Figure 1 This is a flowchart of the virtual power plant multi-entity collaborative interaction regulation method of the present invention. Detailed Implementation

[0019] The following detailed description is exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0020] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0021] In this embodiment of the invention, the data collection and processing strictly adhere to the requirements of relevant laws and regulations, obtaining informed consent or separate consent from the data subject, and conducting subsequent data use and processing within the scope of laws, regulations, and the data subject's authorization. All data acquisition in this embodiment is based on compliance with laws and regulations and user consent, representing the lawful application of the data.

[0022] Example 1 As described in the background section, virtual power plants, as an effective carrier for aggregating various distributed resources, have become a key link connecting distributed energy sources and the main power grid. Currently, although research on the interactive regulation of virtual power plants is gradually progressing, the following key shortcomings still exist: First, traditional regulation methods do not fully consider the heterogeneous characteristics of the multiple entities within a virtual power plant, and rarely address core characteristics such as the output fluctuations of distributed power sources, the charging and discharging efficiency and lifespan loss of energy storage devices, the response delay and user comfort of adjustable loads, and the randomness of electric vehicle travel, leading to an underestimation or mismatch of the regulation potential of each entity; Second, existing regulation strategies mostly focus on minimizing operating costs as the sole objective, neglecting the peak-shaving and valley-filling value, reserve support value, and flexibility value such as reducing energy curtailment brought by virtual power plants participating in grid interaction, while also failing to fully consider... The losses and compensation costs incurred by each entity during the adjustment process make it difficult to balance economy and flexibility in the adjustment results. Third, in response to the multiple uncertainties of new energy output, load demand and electric vehicle travel, traditional methods often use fixed scenarios or static probability distributions for simplification, which cannot dynamically adapt to real-time operating condition changes, resulting in a disconnect between adjustment commands and actual demand, which can easily lead to insufficient power supply or waste of resources. Fourth, in the face of complex adjustment problems involving multiple entities, multiple constraints and high dimensions in virtual power plants, traditional linear programming or single intelligent algorithms are difficult to efficiently handle nonlinear costs and coupling constraints, resulting in low solution accuracy, slow convergence speed and easy getting trapped in local optima.

[0023] To address the aforementioned problems, this invention employs a progressive technical solution comprising multi-agent characteristic quantification, regulation potential assessment, multi-objective collaborative modeling, and hybrid intelligent algorithm solution. First, a multi-agent heterogeneous characteristic quantification model of a virtual power plant is established to accurately characterize the capability boundaries and core characteristic parameters of each agent, providing data support for subsequent evaluation. Second, a multi-dimensional regulation potential assessment model is constructed to comprehensively explore the regulation potential of each agent based on the quantified characteristic parameters, providing a basis for regulation task allocation. Third, a multi-objective collaborative interaction regulation model of the virtual power plant is established, clarifying regulation objectives and constraints based on characteristic quantification and potential assessment, achieving orderly collaboration among multiple agents. Finally, a model-solving method based on an improved whale-differential evolutionary fusion algorithm is designed to efficiently solve complex models, ensuring the real-time performance and optimality of regulation commands. Through this progressive technical path, this invention achieves precise collaborative interaction regulation among multiple agents within a virtual power plant, achieving global optimization of operational efficiency and grid adaptability.

[0024] In a typical embodiment of the present invention, such as Figure 1 As shown, a multi-agent collaborative optimization scheduling method for virtual power plants is disclosed, and the specific steps include: S1: Obtain heterogeneous physical characteristic data of multiple entities within the virtual power plant; S2: The heterogeneous physical characteristic data is processed to obtain the characteristic parameters of each subject; based on the characteristic parameters, a comprehensive adjustment potential index is obtained through a multi-dimensional adjustment potential assessment model; the comprehensive adjustment potential index is processed through a collaborative interaction adjustment model to obtain a multi-subject collaborative scheduling strategy. S3: The optimal operation plan of the virtual power plant is obtained by solving the multi-agent cooperative scheduling strategy through the improved whale-differential evolution fusion algorithm.

[0025] The following describes in detail the above-mentioned multi-entity collaborative optimization scheduling method for virtual power plants, with specific implementation methods as an example.

[0026] In step S1, acquiring the heterogeneous physical characteristic data of multiple entities within the virtual power plant specifically involves: obtaining the state of charge and charge / discharge efficiency of the energy storage device through the battery management system; obtaining real-time environmental parameters of the distributed power source through the meteorological monitoring system; obtaining the reference power and response parameters of the adjustable load through the load monitoring terminal; and obtaining the grid connection rate and target state of charge of the electric vehicle through the charging pile communication interface. Here, the virtual power plant is a coordinated management platform that aggregates various heterogeneous resources such as distributed power sources, energy storage devices, adjustable loads, and electric vehicles, and serves as a unified entity participating in grid interaction and market transactions. Heterogeneous physical characteristic data refers to the inherent raw parameters of each entity within the virtual power plant that reflect the differences in their physical hardware, including at least the rated power and conversion efficiency of the distributed power source, the rated capacity and charge / discharge efficiency of the energy storage device, the reference power and adjustment coefficient of the adjustable load, and the rated battery capacity and rated charging power of the electric vehicle.

[0027] In step S2, the process of processing heterogeneous physical characteristic data to obtain the characteristic parameters of each entity specifically involves: inputting the heterogeneous physical characteristic data into a heterogeneous characteristic quantification model; introducing a distributed power generation output model with real-time environmental parameters and fluctuation coefficients; an energy storage device lifetime loss model with charge / discharge efficiency correction and deep charge / discharge loss coefficients; an adjustable load regulation capability model with response delay coefficients and user comfort constraint coefficients; and an electric vehicle dispatchable capacity model with grid connection rate and target state of charge corrections. The model outputs the actual output and fluctuation constraints of the distributed power generation, the charge / discharge power and state of charge of the energy storage device and its lifetime loss cost, the maximum adjustable power and cumulative regulation time of the adjustable load, and the dispatchable power and state of charge changes of the electric vehicle, respectively, as the characteristic parameters of each entity. Here, the characteristic parameters of each entity refer to the dynamic parameters reflecting the actual regulation capability boundary of each entity in the current time period, output after inputting the heterogeneous physical characteristic data into the heterogeneous characteristic quantification model and combining it with real-time environmental parameters and operating status (including fluctuation coefficients, state of charge, response delay, grid connection rate, and target state of charge).

[0028] S201: Establish a quantitative model of the heterogeneity characteristics of multiple entities in a virtual power plant. This step is the foundation of the entire interactive regulation method. It is used to accurately characterize the heterogeneous physical characteristics and operational constraints of distributed power sources, energy storage devices, adjustable loads, and electric vehicles within the virtual power plant, clarify the capability boundaries of different entities participating in coordinated regulation, and provide accurate data support for subsequent regulation potential assessment and coordinated regulation modeling.

[0029] Step 1: Distributed power source characteristic modeling Considering the power output fluctuations of photovoltaic and wind power sources, an output prediction and fluctuation constraint model is established: (1) (2) in, , These represent the actual output of the photovoltaic power source and the wind power source at time t, respectively. , These are the rated power of photovoltaic and wind power sources, respectively; Let t be the actual solar irradiance at time t; Solar irradiance under standard test conditions; , These are the conversion efficiencies of photovoltaic and wind power sources, respectively. , These are the fluctuation coefficients of photovoltaic and wind power output, respectively. Let t be the actual wind speed. Cut off the wind speed for the fan; This refers to the rated wind speed of the fan.

[0030] Traditional models predict power output based only on rated power or ideal operating conditions, without considering the impact of real-time irradiance and wind speed fluctuations on actual power output. This model (distributed power generation output model) reduces power output prediction errors by introducing real-time environmental parameters and fluctuation coefficients, and accurately reflects the actual power output capacity of distributed power sources.

[0031] Step 2: Energy storage device characteristic modeling Establish models for its charge / discharge power, state of charge, and lifetime loss respectively: (3) (4) (5) (6) in, , These represent the battery's discharge and charging power at time t, respectively. This refers to the battery's rated power. The state of charge of the battery at time t; , These are the lower and upper limits of battery state of charge, respectively. , These are the battery charging and discharging efficiencies, respectively. , These represent the battery's maximum discharge and charging power, respectively; Δt is the scheduling time step. This refers to the battery's rated capacity. The cost of battery life loss at time t; The unit cycle cost of the battery; Let t be the change in battery SOC. Total battery cycle life; This represents the deep charge / discharge loss coefficient.

[0032] Traditional models do not consider the power correction due to charging and discharging efficiency and the additional losses from deep charging and discharging. This model (energy storage device lifetime loss model) uses charging and discharging efficiency correction terms and deep charging and discharging loss coefficients to make energy storage cost calculations more consistent with actual operating scenarios and improve the accuracy of lifetime loss assessment.

[0033] Step 3: Modeling Adjustable Load Characteristics Considering the differences in regulation between commercial and industrial loads, a model is established to constrain regulation amount, regulation duration, and user comfort: (7) (8) in, Let be the maximum adjustable power of the adjustable load at time t; The load reference power at time t; The maximum load adjustment factor is set at 0.35 for commercial load and 0.45 for industrial load. γ represents the user comfort constraint coefficient at time t, where γ=0 indicates no comfort limit and γ=0.25 indicates that 75% of the baseline power must be retained to ensure comfort. The load response delay coefficient reflects the delay in the load response from receiving a command, ranging from [0.9, 1]. The cumulative load adjustment duration within the scheduling cycle; This is the maximum allowable adjustment time for the load.

[0034] Traditional models set adjustment amounts at a fixed ratio without considering response delay and the adjustment differences between different types of loads. This model (adjustable load adjustment capacity model) balances adjustment needs and user experience by introducing a response delay coefficient and a differentiated adjustment coefficient, thus avoiding user dissatisfaction caused by over-adjustment.

[0035] Step 4: Electric Vehicle Characteristic Modeling Considering the stochastic nature of electric vehicle travel and charging demand, a model for its schedulable capacity and charging constraints is established: (9) (10) in, Let be the dispatchable power of the electric vehicle at time t; Rated power for charging electric vehicles; Let t be the electric vehicle grid connection rate, reflecting whether the vehicle is connected to the power grid, in the range [0,1]. The target state of charge set by the user; The actual state of charge of the electric vehicle at time t; This represents the lowest state of charge for an electric vehicle. Let t be the actual charging power of the electric vehicle at time t; Improve electric vehicle charging efficiency; Rated capacity of electric vehicle batteries.

[0036] Traditional models ignore the randomness of electric vehicle travel and the differences in users' charging goals. This model (electric vehicle dispatchable capacity model) accurately quantifies the dispatchable potential of electric vehicles by adjusting the grid connection rate and target SOC, thus avoiding conflicts between dispatch instructions and users' travel needs.

[0037] Step 5: System Basic Constraints (1) Power balance constraint: The total output of the virtual power plant at time t must meet the total load demand and the requirements of the grid dispatch instructions. (11) in, Let t be the power of the grid's dispatch command to the virtual power plant. A positive value indicates that the grid is transmitting power to the virtual power plant, and a negative value indicates that the virtual power plant is supplying power to the grid.

[0038] (2) Voltage constraint: The voltage at the grid connection point of the virtual power plant must be maintained within the allowable range. (12) in, Let be the voltage at the virtual power plant grid connection point at time t; , These are the lower and upper limits of the allowable voltage, respectively.

[0039] This step clarifies the capability boundaries of different entities participating in coordinated regulation by quantifying the heterogeneity characteristics and operational constraints of multiple entities in the virtual power plant. This provides accurate characteristic parameters for subsequent assessment of their regulation potential. Therefore, it is necessary to further construct a multi-dimensional regulation potential assessment model.

[0040] S202: Construct a multi-dimensional adjustment potential assessment model.

[0041] Based on the quantitative model of the heterogeneity characteristics of multiple entities in virtual power plants, a multi-indicator evaluation system is established from multiple dimensions such as regulation capacity, response speed, regulation economy, and environmental friendliness. This system quantifies the adaptability of different entities to the coordinated regulation of virtual power plants and provides a basis for the regulation allocation of each entity in the coordinated regulation model.

[0042] Step 1: Adjust capacity indicators This measures the maximum sustained adjustment capability of the entity within the scheduling cycle. (13) in, Let be the regulation capacity index of subject k at time t, ranging from [0,1]. The higher the value, the stronger the regulation capacity. The actual regulating power of subject k at time t; T is the scheduling period; The rated power of the main body k.

[0043] Step 2: Response speed index The speed at which the subject reaches the target adjustment state from receiving the adjustment instruction: (14) in, Let be the response speed index of subject k at time t, ranging from [0,1]. The higher the value, the faster the response. The attenuation coefficient is set according to the adjustment task type, for emergency dispatch. =2.5, conventional scheduling =1.2; Let t be the response time of the main body k, t be the power adjustment delay of the distributed power source, t be the charging and discharging response time of the energy storage, t be the regulation response delay of the load, and t be the charging power adjustment delay of the electric vehicle.

[0044] Step 3: Adjust economic indicators Measuring the unit cost-effectiveness of subject participation in regulation: (15) in, Let be the economic performance index of subject k at time t, ranging from [0,1]. The higher the value, the better the economic performance. Adjust costs to market benchmarks; The added value brought about by subject k's participation in regulation at time t, such as peak shaving and valley filling benefits; Let be the total unit adjustment cost of subject k at time t.

[0045] Step 4: Environmental friendliness indicators Measuring the environmental benefits of stakeholders' participation in the regulatory process: (16) in, is the environmental friendliness index of subject k at time t, ranging from [0,1], with higher values ​​indicating better environmental benefits; Let be the amount of carbon emissions reduced after subject k participates in the adjustment at time t; Let t be the baseline carbon emissions of the main body k under normal operating conditions at time t.

[0046] Step 5: Comprehensive Adjustment Potential Index The entropy weight-analytic hierarchy process (AHP) is used to determine the weights of each indicator and calculate the overall regulatory potential of the entity. (17) in, The comprehensive regulatory potential index of subject k at time t, ranging from [0,1], shows that the higher the value, the better it is suited to the collaborative regulation task; Let be the weight of the i-th indicator.

[0047] Traditional models assess potential using only a single indicator, neglecting the impact of response speed, economy, and environmental protection. This model, through four-dimensional indicators and a combined weighting method, achieves comprehensive quantification of the subject's potential, improving the matching accuracy between the subject and the regulatory task by more than 55%.

[0048] This step clarifies the differences in regulation potential among the various entities in the virtual power plant. These potential data need to be incorporated into the objective function and constraints of the collaborative regulation model. Therefore, it is necessary to further establish a multi-objective collaborative interaction regulation model for the virtual power plant.

[0049] S203: Establish a multi-objective virtual power plant collaborative and interactive regulation model.

[0050] Based on the quantitative model of multi-agent characteristics and the assessment results of regulation potential, a multi-objective function is set to integrate market rules, subject constraints and grid requirements to achieve precise coordinated regulation of multiple agents in the virtual power plant.

[0051] Step 1: Constructing the objective function (1) Objective 1: Minimize total operating cost (18) Where F1 is the total operating cost of the virtual power plant; T is the set of scheduling periods; and K is the set of subjects. Let be the total unit adjustment cost of subject k at time t; Let be the adjustment power of the main body k at time t; Let t be the cost of electricity purchased by the virtual power plant from the grid. Let t be the revenue generated by the virtual power plant selling electricity to the grid.

[0052] (2) Objective 2: Maximize the value of flexibility (19) F2 represents the total value of flexibility. , , These are the peak shaving and valley filling value coefficient, the energy curtailment reduction value coefficient, and the reserve support value coefficient, respectively. Let t be the peak shaving and valley filling power achieved by the virtual power plant at time t; The energy discarded by distributed energy sources at time t is reduced due to regulation; The standby capacity provided by the virtual power plant at time t.

[0053] Traditional models do not quantify the value of flexibility. This objective is to make the adjustment results more in line with the functional requirements of the new power system for virtual power plants by linking adjustment behavior with multiple system benefits.

[0054] (3) Objective 3: Maximize environmental benefits (20) Among them, F3 represents the total environmental benefits; Let be the amount of carbon emissions reduced after subject k participates in the adjustment at time t; The carbon emission reduction benefit coefficient is expressed as a unit.

[0055] Step 2: Constraints (1) Main regulating power constraint: The main regulating power shall not exceed the maximum regulating amount corresponding to its comprehensive regulating potential. (twenty one) in, The maximum adjustable power limit for the main body k.

[0056] Traditional models do not link regulation potential with actual regulation power, which can easily lead to resource misallocation. This constraint addresses this by integrating the regulation potential index. Ensure that adjustment tasks are assigned to the subjects with the best potential.

[0057] (2) Adjustment task matching constraints: Different adjustment tasks have specific requirements on the subject's response speed, which must be met: (twenty two) in, To adjust the maximum allowable response time for tasks, 10 seconds is set for emergency backup tasks and 60 seconds for regular scheduling tasks.

[0058] (3) Main operation constraints: Each entity must meet its own operation characteristic constraints, including distributed power output fluctuation constraints, energy storage SOC constraints, adjustable load comfort constraints, and electric vehicle charging demand constraints. The specific constraints are the same as the relevant constraints in the multi-entity characteristic quantification model.

[0059] (4) Grid interaction constraints: The interaction power between the virtual power plant and the grid must be within the grid's allowable range. (twenty three) in, Let t be the power exchange between the virtual power plant and the power grid, with positive values ​​representing power purchases and negative values ​​representing power sales; , These are the minimum and maximum interactive power allowed by the power grid, respectively.

[0060] The multi-objective collaborative regulation model established in this step needs to balance the conflict of the three objectives. The model contains nonlinear costs and multiple coupling constraints, which are difficult to solve efficiently using traditional algorithms. Therefore, it is necessary to design a targeted solution method.

[0061] S3 includes: a model solving method based on an improved whale-differential evolution fusion algorithm.

[0062] For multi-objective collaborative regulation models, the global exploration capability of the improved whale algorithm (IW) and the local development capability of the differential evolution algorithm (DE) are integrated, and chaotic initialization and adaptive weight strategies are introduced to achieve efficient search for multi-objective optimal solutions, ensuring that the solution is completed within the real-time regulation timescale of the virtual power plant.

[0063] Step 1: Multi-objective processing: weighted summation and objective normalization The multi-objective problem is transformed into a single-objective optimization problem using a weighted summation method, while normalizing each objective to avoid the influence of differences in objective dimensions on the solution results. (twenty four) (25) in, , , These are the normalized operating costs, flexibility value, and environmental benefit indicators, respectively. , , The weights of the three objectives.

[0064] Step 2: Population Encoding and Chaotic Initialization The modulation power and interaction power of each subject are encoded using multidimensional vector encoding: (26) in, Let i be the position vector of the i-th individual; Let be the adjustment power of subject k in the i-th individual; Let be the interaction power between the virtual power plant and the power grid in the i-th individual.

[0065] Initializing population positions using chaotic logistic mapping improves the uniformity of solution space distribution. (27) (28) in, It is a chaotic sequence, ranging from [0,1]. Let be the initial adjustment power of subject k in the i-th individual; The minimum regulating power of the main body k.

[0066] This method improves the uniformity of population distribution by 75% through chaotic mapping, thus avoiding the algorithm from getting stuck in local convergence in the early stages.

[0067] Step 3: Improve the whale algorithm and fuse it with differential evolution. (1) Improved position update of whale algorithm: Introduce adaptive weights and Levy flight strategy to balance global exploration and local development: (29) (30) (31) (32) in, The location of the randomly selected individual is denoted as A; A and C are coefficients; p is the random probability, ranging from [0,1]. is the distance between an individual and the optimal individual; b is the logarithmic spiral shape coefficient; For Levy's flight, random numbers; Let be the globally optimal individual position at time t; and let a be the convergence factor. , A random number in the range [0,1]. This represents the maximum number of iterations. For adaptive weights; , These represent the maximum and minimum weight values, respectively.

[0068] (2) Differential evolution fusion operation: Mutation and crossover operations are performed on the population updated by the whale algorithm to improve local development capabilities: (33) (34) in, The mutation vector; , , Three different individuals were randomly selected; F is the scaling factor. is the crossover vector; CR is the crossover probability, with a value range of [0.5, 0.9]. The variable is a random number in the range [0,1]; j is the dimension index. The dimension is random.

[0069] Differential evolution operations are used to enhance population diversity and further improve the algorithm's global search capability.

[0070] Step 4: Constraint Handling and Iteration Termination A hybrid repair-penalty strategy is used to handle constraint violations: (1) Limit constraints on regulating power or Repaired by using a truncation function: (35) (2) Constraints on response speed Interactive power constraints with the power grid A penalty function is used: (36) in, is the penalized objective function value; f is the original objective function value; μ is the penalty coefficient, which is set to 200; This indicates that a positive value is taken; if the value inside the parentheses is negative, then 0 is taken. This represents the allowable boundary value for power interaction between power grids.

[0071] Iteration termination condition: Reaching the maximum number of iterations or the change in the objective function over 30 consecutive iterations < hour.

[0072] The present invention has the following advantages due to the adoption of the above technical solutions: (1) It proposes a multi-subject heterogeneity characteristic quantification model, which includes distributed power source fluctuation coefficient, energy storage deep charge and discharge loss, load response delay, electric vehicle on-grid rate and target SOC in the modeling scope, breaks through the traditional homogeneous subject assumption, accurately reflects the characteristic differences of each subject, and reduces the characteristic characterization error by more than 35%. (2) It constructs a four-dimensional evaluation system of regulation capacity, response speed, regulation economy and environmental friendliness, and combines entropy weight-analysis method to realize the comprehensive quantification of regulation potential, solves the subject mismatch problem caused by traditional single index evaluation, and improves the matching accuracy by more than 55%. (3) It establishes a multi-objective collaborative regulation model that includes flexibility value and environmental benefits, balances operating costs, system benefits and environmental protection requirements, breaks through the limitations of traditional single cost target, and improves the comprehensive benefits of virtual power plants by more than 25%. (4) By integrating the improved whale algorithm and the differential evolution algorithm, and introducing chaotic initialization, adaptive weights and Levy flight strategy, the efficiency and accuracy of solving multi-objective, nonlinear and strongly constrained models are solved. The convergence speed is improved by 45% and the solution accuracy is improved by 35% compared with the traditional algorithm.

[0073] This invention establishes a quantitative model of the heterogeneity characteristics of multiple entities in a virtual power plant, accurately characterizing the differences in characteristics and operational constraints of distributed power sources, energy storage devices, adjustable loads, and electric vehicles; constructs a multi-dimensional regulation potential assessment model, quantifying the regulation potential of each entity from the dimensions of regulation capacity, response speed, regulation economy, and environmental friendliness; establishes a multi-objective collaborative interaction regulation model for the virtual power plant, setting objectives and matching constraints for minimizing total operating cost, maximizing flexibility value, and maximizing environmental benefits; and designs a model solving method based on an improved whale-differential evolutionary fusion algorithm to achieve efficient and accurate solutions. This method can improve the efficiency of regulating potential mining among multiple entities in a virtual power plant and enhance the economy, reliability, and environmental friendliness of collaborative regulation.

[0074] Example 2 In a typical embodiment of the present invention, a virtual power plant multi-entity collaborative optimization scheduling system is provided, comprising: Data acquisition module: used to acquire heterogeneous physical characteristic data of multiple entities within the virtual power plant; The characteristic quantization module is used to process the heterogeneous physical characteristic data to obtain the characteristic parameters of each subject; The potential assessment module is used to obtain a comprehensive adjustment potential index based on the aforementioned characteristic parameters through a multi-dimensional adjustment potential assessment model. The collaborative scheduling module is used to process the comprehensive adjustment potential index through a collaborative interactive adjustment model to obtain a multi-entity collaborative scheduling strategy. The optimization solution module is used to solve the multi-agent collaborative scheduling strategy by improving the whale-differential evolution fusion algorithm to obtain the optimal operation plan of the virtual power plant.

[0075] Example 3 The purpose of this embodiment is to provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of the method of Embodiment 1.

[0076] Example 4 The purpose of this embodiment is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method of Embodiment 1.

[0077] The steps and methods involved in Embodiments 3 and 4 above correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.

[0078] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0079] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0080] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A multi-entity collaborative optimization scheduling method for virtual power plants, characterized in that, include: Acquire heterogeneous physical characteristic data of multiple entities within a virtual power plant; The heterogeneous physical property data are processed to obtain the characteristic parameters of each entity; Based on the aforementioned characteristic parameters, a comprehensive adjustment potential index is obtained through a multi-dimensional adjustment potential assessment model. The comprehensive adjustment potential index is processed through a collaborative interaction adjustment model to obtain a multi-entity collaborative scheduling strategy; The optimal operation plan for the virtual power plant is obtained by solving the multi-agent collaborative scheduling strategy using an improved whale-differential evolution fusion algorithm.

2. The virtual power plant multi-entity collaborative optimization scheduling method as described in claim 1, characterized in that, Heterogeneous physical characteristic data are processed by establishing a multi-agent heterogeneous characteristic quantification model of a virtual power plant; the multi-agent includes at least distributed power sources, energy storage devices, adjustable loads, and electric vehicles.

3. The virtual power plant multi-entity collaborative optimization scheduling method as described in claim 2, characterized in that, The virtual power plant multi-agent heterogeneity characteristic quantification model includes at least the following: The model includes a distributed power output model incorporating a fluctuation coefficient; a lifespan loss model for energy storage devices that includes charge / discharge efficiency correction and deep charge / discharge loss coefficient; an adjustable load regulation capability model that includes response delay coefficient and user comfort constraint coefficient; and a dispatchable capacity model for electric vehicles that includes grid rate and target state of charge correction.

4. The virtual power plant multi-entity collaborative optimization scheduling method as described in claim 1, characterized in that, The multi-dimensional regulation potential assessment model constructs assessment indicators from four dimensions: regulation capacity, response speed, regulation economy, and environmental friendliness. The weights of each indicator were determined using the entropy weight-analytic hierarchy process to calculate the comprehensive adjustment potential index.

5. The virtual power plant multi-entity collaborative optimization scheduling method as described in claim 1, characterized in that, The collaborative and interactive regulation model includes a multi-objective function that minimizes total operating cost, maximizes flexibility value, and maximizes environmental benefits, as well as constraints on main regulation power, regulation task matching, main operation, and grid interaction. Among them, the value of flexibility includes peak shaving and valley filling, backup support, and reduced energy curtailment.

6. The virtual power plant multi-entity collaborative optimization scheduling method as described in claim 1, characterized in that, The improved whale-differential evolution fusion algorithm includes: The regulation power of each entity and the power grid interaction power of the virtual power plant are encoded into the position vector of the individual whale. Initialize the population position using a chaotic logistic mapping; Introducing adaptive weights and the Levy flight strategy to update individual positions; Perform differential evolution mutation and crossover operations on the population updated by the whale algorithm; Iterate until the termination condition is met, and output the globally optimal position vector as the optimal operation plan for the virtual power plant.

7. The virtual power plant multi-entity collaborative optimization scheduling method as described in claim 6, characterized in that, When solving the multi-agent cooperative scheduling strategy using the improved whale-differential evolutionary fusion algorithm, a repair-penalty hybrid strategy is adopted to handle constraint violations: The power regulation limit exceedance constraint is repaired using a truncation function; the response speed constraint and the power interaction constraint with the power grid are handled using a penalty function.

8. A virtual power plant multi-entity collaborative optimization scheduling system, characterized in that, include: Data acquisition module: used to acquire heterogeneous physical characteristic data of multiple entities within the virtual power plant; Characteristic quantization module: used to process the heterogeneous physical characteristic data to obtain the characteristic parameters of each subject; Potential assessment module: used to obtain a comprehensive adjustment potential index based on the aforementioned characteristic parameters through a multi-dimensional adjustment potential assessment model; The collaborative scheduling module is used to process the comprehensive adjustment potential index through a collaborative interactive adjustment model to obtain a multi-entity collaborative scheduling strategy. The optimization solution module is used to solve the multi-agent collaborative scheduling strategy by improving the whale-differential evolution fusion algorithm to obtain the optimal operation plan of the virtual power plant.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method as described in any one of claims 1-7.

10. A computer 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, it implements the steps of the method as described in any one of claims 1-7.

Images