A virtual power plant dispatch optimization method and system

Abstract

This invention provides a virtual power plant scheduling optimization method and system. The method includes: constructing an evolutionary strategy model for the virtual power plant with the objective functions of maximizing economic benefits, minimizing power deviation, and minimizing operating costs; collecting real-time operating data of each distributed energy source in the virtual power plant; using the preprocessed real-time operating data and environmental conditions as model inputs to the evolutionary strategy model, training and optimizing the scheduling strategy using an evolutionary strategy algorithm, and outputting the optimal scheduling strategy for the virtual power plant; and using the optimized scheduling strategy to regulate each distributed energy source in the virtual power plant. This invention generates and evaluates scheduling strategies by simulating population evolution, enabling the rapid finding of near-optimal solutions in complex, multi-objective virtual power plant scheduling scenarios. It also exhibits strong robustness and flexibility, adapting to different electricity market environments and scheduling needs.

Description

Technical Field

[0001] This invention relates to the field of smart grid dispatch optimization technology, specifically to a virtual power plant dispatch optimization method and system. Background Technology

[0002] A virtual power plant (VPP) is a system that integrates and coordinates the control of distributed energy resources (such as photovoltaic, wind power, and energy storage) through information and communication technologies. VPPs achieve efficient utilization of power resources through optimized scheduling and participate in electricity market transactions.

[0003] Currently, traditional virtual power plant (VPP) scheduling methods (such as linear programming and dynamic programming) are inefficient when facing complex nonlinear, multi-objective optimization problems, and struggle to cope with the highly uncertain renewable energy output and diverse user demands in VPPs. Furthermore, while deep reinforcement learning methods based on machine learning perform well, they often require large amounts of training data and computational resources, making rapid deployment in real-world systems difficult. In addition, existing VPP scheduling methods also suffer from the following major drawbacks:

[0004] (1) Strong data dependence: Existing deep reinforcement learning algorithms require a large amount of data for training and are prone to failure when faced with new scenarios or abnormal situations, lacking flexibility;

[0005] (2) High computational resource consumption: Deep learning algorithms usually require a lot of computational resources for training and inference, which poses a challenge to the real-time requirements of virtual power plant scheduling optimization.

[0006] (3) Slow algorithm convergence: Existing optimization methods converge slowly in complex scheduling scenarios, making it difficult to provide high-quality scheduling solutions in a short time. Summary of the Invention

[0007] Therefore, this invention provides a virtual power plant scheduling optimization method and system, aiming to solve the technical problem that existing technologies are insufficient in dealing with complex nonlinear and multi-objective optimization problems, and that it is difficult to quickly and accurately deploy virtual power plant scheduling optimization.

[0008] To achieve the above objectives, the present invention adopts the following technical solution:

[0009] According to a first aspect of the present invention, the present invention provides a virtual power plant scheduling optimization method, the method comprising:

[0010] For the virtual power plant, an evolutionary strategy model is constructed with the objective functions of maximizing economic benefits, minimizing power deviation, and minimizing operating costs.

[0011] Collect real-time operating data of each distributed energy source in the virtual power plant;

[0012] The preprocessed real-time operating data and system status are used as model inputs to the evolutionary strategy model. The evolutionary strategy algorithm is used to train and optimize the scheduling strategy, and the optimal scheduling strategy for the virtual power plant is output.

[0013] The scheduling optimization strategy is used to regulate each distributed energy source in the virtual power plant;

[0014] The real-time operating data includes at least one of the following: output power, load demand, energy storage status, and electricity price.

[0015] Optionally, the objective function formula is expressed as follows:

[0016]

[0017] Wherein, revenue t is the revenue calculated based on electricity price and electricity trading volume, in yuan; power deviation t The deviation between actual power supply and demand, in MW; operating costs. t This includes energy losses from charging and discharging of energy storage devices and operation and maintenance costs, expressed in yuan; weighting coefficients α, β, and λ represent revenue, respectively. t Power deviation t The weights of the operating cost t are used to adjust the priority of the optimization objectives; the strategy parameter θ represents the scheduling strategy for the virtual power plant.

[0018] Optionally, the step of training and optimizing the scheduling strategy using an evolutionary strategy algorithm to output the optimal scheduling strategy for the virtual power plant includes:

[0019] Multiple sets of initial policy parameters are randomly generated to form an initial population;

[0020] The fitness of each set of initial policy parameters in the initial population is evaluated using the objective function.

[0021] Select initial policy parameters with high fitness for mutation operation, and perform population iteration;

[0022] During the population iteration process, the strategy parameters that meet the fitness criteria are selected as the optimal scheduling strategy for the virtual power plant.

[0023] Optionally, the random generation of multiple sets of initial policy parameters to form an initial population includes:

[0024] Set the initial population size to N, and generate N sets of random policy parameters θ at the initial time. i , i = 1, 2, ..., N;

[0025] The initial population formula is expressed as follows:

[0026]

[0027] Optionally, evaluating the fitness of each set of initial policy parameters in the initial population using the objective function includes:

[0028] Calculating the objective function value of each initial policy parameter group in the initial population as the fitness specifically includes:

[0029] For each time t, based on the policy parameters and the system state s of the virtual power plant t Generate the corresponding system action a t ;

[0030] Based on the system state s t and the system action a t The system action a is executed. t The target system state s t+1 ;

[0031] The objective function value is calculated based on the state of the target system.

[0032] Optionally, the system state s t The formula is expressed as follows:

[0033]

[0034] in, The output power of the distributed energy source at time t, in MW; The load demand at time t, in MW; S t The energy storage state at time t, expressed in MWh; M t The price of electricity at time t is expressed in yuan / MWh.

[0035] The system action a t The formula is expressed as follows:

[0036]

[0037] in, For policy functions; The charging power of the energy storage device at time t, expressed in MW; The discharge power of the energy storage device at time t, expressed in MW; The output power of the distributed energy source at time t is expressed in MW.

[0038] The target system state s t+1 The formula is expressed as follows:

[0039]

[0040] in, The output power of the distributed energy source at time t+1 is expressed in MW. The load demand at time t+1 is expressed in MW; S t+1 The energy storage state at time t+1 is expressed in MWh; M t+1 The price is the electricity price at time t+1, expressed in yuan / MWh.

[0041] Optionally, the step of selecting initial policy parameters with higher fitness for mutation operation and performing population iteration includes:

[0042] Step S1: Select the k strategy parameters with high fitness using either the roulette wheel selection method or the tournament selection method;

[0043] Step S2: Add random noise to the k policy parameters to obtain the next generation of policy parameters, expressed by the following formula:

[0044]

[0045] in, σ represents the next-generation strategy parameters; -(0,I) represents a normal distribution with a mean of 0 and a variance equal to the identity matrix.

[0046] Step S3: Replace the initial population with the new generation strategy parameters to form a new generation population. And repeat steps S1 to S2.

[0047] Optionally, selecting the strategy parameters that meet the fitness criteria as the optimal scheduling strategy for the virtual power plant during the population iteration process includes:

[0048] During the population iteration process, the fitness of each set of strategy parameters is calculated in real time. If the change in fitness is less than a preset threshold, or the maximum number of iterations is reached, the population iteration is terminated.

[0049] After the population iteration terminates, the strategy parameter with the highest fitness in the latest population is output as the optimal scheduling strategy for the virtual power plant.

[0050] Optionally, the step of regulating each distributed energy source in the virtual power plant using the scheduling optimization strategy includes:

[0051] The target system action is generated based on the optimal scheduling strategy, as expressed by the following formula:

[0052]

[0053] in, Actions for the target system; For policy functions; These are the policy parameters corresponding to the optimal scheduling policy;

[0054] The charging power, discharging power, and / or output power of the energy storage devices corresponding to each distributed energy source in the virtual power plant are regulated by the actions of the target system.

[0055] According to a second aspect of the present invention, the present invention provides a virtual power plant dispatch optimization system, the system comprising:

[0056] The model building module is used to construct an evolutionary strategy model for the virtual power plant with the objective functions of maximizing economic benefits, minimizing power deviation, and minimizing operating costs.

[0057] The data acquisition module is used to collect real-time operating data of each distributed energy source in the virtual power plant;

[0058] The strategy prediction module is used to take the preprocessed real-time operating data and system status as the model input of the evolutionary strategy model, use the evolutionary strategy algorithm to train and optimize the scheduling strategy, and output the optimal scheduling strategy for the virtual power plant.

[0059] An optimization execution module is used to regulate each distributed energy source in the virtual power plant using the scheduling optimization strategy;

[0060] The real-time operating data includes at least one of the following: output power, load demand, energy storage status, and electricity price.

[0061] The present invention, by adopting the above technical solution, has at least the following beneficial effects:

[0062] The present invention constructs an evolutionary strategy model for the virtual power plant, with the objective functions of maximizing economic benefits, minimizing power deviation, and minimizing operating costs. Real-time operating data of each distributed energy source in the virtual power plant is collected. The preprocessed real-time operating data and system status are used as inputs to the evolutionary strategy model. An evolutionary strategy algorithm is used to train and optimize the scheduling strategy, outputting the optimal scheduling strategy for the virtual power plant. The optimized scheduling strategy is then used to regulate each distributed energy source in the virtual power plant. Thus, by generating and evaluating scheduling strategies through simulated population evolution, a near-optimal solution can be quickly found in complex, multi-objective virtual power plant scheduling scenarios. This approach exhibits strong robustness and flexibility, adapting to different electricity market environments and scheduling needs.

[0063] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0064] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0065] Figure 1 A flowchart illustrating a virtual power plant scheduling optimization method according to an embodiment of the present invention is shown.

[0066] Figure 2 A schematic diagram of the structure of a virtual power plant dispatch optimization system provided in an embodiment of the present invention is shown;

[0067] Figure 3 A schematic diagram of the physical structure of a computer device provided in an embodiment of the present invention is shown. Detailed Implementation

[0068] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0069] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0070] This invention provides a virtual power plant scheduling optimization method, such as... Figure 1 As shown, it may include at least the following steps S101 to S104:

[0071] Step S101: For the virtual power plant, construct an evolutionary strategy model with the objective functions of maximizing economic benefits, minimizing power deviation, and minimizing operating costs.

[0072] Evolutionary strategy is an optimization method based on natural selection and genetic algorithms. It optimizes the objective function by simulating the evolutionary process of a population, offering advantages such as strong adaptability and high computational efficiency. It does not rely on gradient information and is suitable for high-dimensional, non-convex optimization problems. In this embodiment of the invention, an evolutionary strategy model is used to perform scheduling optimization analysis on various distributed energy sources in a virtual power plant. The objective function is defined as the cumulative revenue over the entire scheduling period from t=1 to T, with the optimization objective being optimal economic benefit, including maximizing revenue, minimizing power deviation, and reducing operating costs. The objective function is defined as the cumulative revenue over the entire scheduling period from t=1 to T, expressed by the following formula:

[0073]

[0074] Wherein, revenue t is the revenue calculated based on electricity price and electricity trading volume, in yuan; power deviation t The deviation between actual power supply and demand, in MW; operating costs. t This includes energy losses from charging and discharging of energy storage devices and operation and maintenance costs, expressed in yuan; weighting coefficients α, β, and λ represent revenue, respectively. t Power deviation t Operating costs t The weights are used to adjust the priority of the optimization objectives; the strategy parameter θ is a set of parameters used to generate system actions, representing the scheduling strategy for the virtual power plant.

[0075] The evolutionary strategy model proposed in this invention is based on an evolutionary strategy algorithm to train and optimize the scheduling strategy. The model input includes preprocessed real-time running data and system state, and the output is the scheduling decision.

[0076] Step S102: Collect real-time operating data of each distributed energy source in the virtual power plant.

[0077] Real-time operational data includes, but is not limited to, power generation, load demand, and energy storage status, used to determine the dispatchability of each distributed energy source in the virtual power plant. In evolutionary strategy-based virtual power plant scheduling optimization, the system state and related parameters of the virtual power plant must first be defined.

[0078] System status s t It can be represented as a vector:

[0079]

[0080] in, The output power of distributed energy sources (such as photovoltaics and wind power) at time t, expressed in MW; The load demand at time t, in MW; S t The energy storage (e.g., energy level) state at time t, expressed in MWh; Mt The price of electricity at time t is expressed in yuan / MWh.

[0081] System action a t It can be represented as a vector:

[0082]

[0083] in, The charging power of the energy storage device at time t, expressed in MW; The discharge power of the energy storage device at time t, expressed in MW; The output power of the distributed energy source at time t is expressed in MW.

[0084] Step S103: The preprocessed real-time operating data and system status are used as model inputs to the evolutionary strategy model. The evolutionary strategy algorithm is used to train and optimize the scheduling strategy, and the optimal scheduling strategy for the virtual power plant is output.

[0085] Specifically, the evolutionary strategy algorithm may include randomly generating multiple sets of initial strategy parameters to form an initial population; evaluating the fitness of each set of initial strategy parameters in the initial population using an objective function; selecting initial strategy parameters with higher fitness for mutation operations and performing population iteration; and selecting strategy parameters with satisfactory fitness as the optimal scheduling strategy for the virtual power plant during the population iteration process.

[0086] In other words, we first need to set the initial population size to N, and then generate N sets of random policy parameters θ at the initial time. i , i = 1, 2, ..., N;

[0087] The initial population formula is expressed as follows:

[0088]

[0089] Furthermore, for each policy parameter in the initial population The evaluation involves calculating the objective function value as the fitness, which may specifically include: for each time t, based on the policy parameters... and the system status of the virtual power plant t Generate the corresponding system action a t ;

[0090]

[0091] in, This is the policy function, whose purpose is to determine the current system state s. t and strategy parameters Generate the action a at the current time t. tIn reinforcement learning, the policy function is a deterministic function, which obtains the actions of the virtual power plant by inputting the policy.

[0092] Therefore, the system state s can be utilized. t and system action a t This allows us to obtain the result of executing system action a. t The subsequent target system state s t+1 :

[0093]

[0094] in, The output power of the distributed energy source at time t+1 is expressed in MW. The load demand at time t+1 is expressed in MW; S t+1 The energy storage state at time t+1 is expressed in MWh; M t+1 The price is the electricity price at time t+1, expressed in yuan / MWh.

[0095] Furthermore, initial policy parameters with high fitness are selected for mutation operations, and population iteration is performed, which may specifically include the following steps S1 to S3:

[0096] Step S1: Select the k strategy parameters with high fitness using either Roulette Wheel Selection or Tournament Selection.

[0097] Step S2: Add random noise to the k policy parameters to obtain the next generation of policy parameters, expressed by the following formula:

[0098]

[0099] in, ε is the parameter of the next-generation strategy; σ is the mutation intensity; ε(0,I) is a normal distribution with a mean of 0 and a variance of the identity matrix;

[0100] Step S3: Replace the initial population with the new generation of policy parameters to form a new generation population. And repeat steps S1 to S2.

[0101] In other words, the new generation of population is replaced with the new generation of policy parameters after noise is added to form a new generation of population, and the policy evaluation and selection mutation process continues until the stopping condition is met.

[0102] Furthermore, during the population iteration process, the fitness of each set of policy parameters can be calculated in real time. If the fitness change is less than a preset threshold, or if the maximum number of iterations G is reached, the population can proceed. maxIf the population iteration fails, it is considered to have converged to the optimal solution, and the population iteration is terminated. After the population iteration terminates, the policy parameter with the highest fitness in the latest population is output. As the optimal scheduling strategy for virtual power plants, it can be used for scheduling execution in actual systems.

[0103] Step S104: Use scheduling optimization strategies to regulate each distributed energy source in the virtual power plant.

[0104] In actual system operation, according to the optimal scheduling strategy The target system action is generated, expressed by the following formula:

[0105]

[0106] in, Actions for the target system; For policy functions; These are the policy parameters corresponding to the optimal scheduling policy. Furthermore, the actions of the target system are utilized. The charging power, discharging power, and / or output power of the energy storage devices corresponding to each distributed energy source in the virtual power plant are regulated.

[0107] This invention provides a virtual power plant scheduling optimization method. It constructs an evolutionary strategy model for the virtual power plant, with the objective functions of maximizing economic benefits, minimizing power deviation, and minimizing operating costs. Real-time operating data of each distributed energy source in the virtual power plant is collected. The preprocessed real-time operating data and system state are used as inputs to the evolutionary strategy model. An evolutionary strategy algorithm is used to train and optimize the scheduling strategy, outputting the optimal scheduling strategy for the virtual power plant. The optimized scheduling strategy is then used to regulate each distributed energy source in the virtual power plant. This invention addresses the shortcomings of traditional scheduling methods in handling complex nonlinear, multi-objective optimization problems by proposing an innovative solution. By simulating natural selection and genetic algorithm evolutionary strategies, this invention can quickly adapt to and optimize the scheduling of distributed energy sources (such as photovoltaic, wind power, and energy storage) in a virtual power plant without relying on large amounts of data and computational resources, thereby maximizing economic benefits, minimizing power deviation, and minimizing operating costs, while simultaneously improving the system's robustness and flexibility. Furthermore, this technology also involves the fields of electricity market trading and real-time operational data processing. By collecting and processing key data such as power generation, load demand, and energy storage status of distributed energy resources in real time, it provides input to the evolutionary strategy model, thereby generating and executing scheduling decisions to achieve efficient utilization of electricity resources and optimization of market transactions. In summary, this invention utilizes an evolutionary strategy to optimize virtual power plant scheduling, improving the system's economic efficiency and operational stability. This method does not rely on gradient information, has high computational efficiency, and is suitable for complex and ever-changing electricity market environments. The technical fields directly applied by this invention cover multiple aspects such as smart grid scheduling optimization, virtual power plant management, electricity market trading strategies, and real-time data processing and analysis, providing important technical support and innovative ideas for the future development of virtual power plant scheduling management.

[0108] Furthermore, as Figure 1 In specific implementation, embodiments of the present invention provide a virtual power plant scheduling optimization system, such as... Figure 2 As shown, the system may include: a model building module 210, a data acquisition module 220, a strategy prediction module 230, and an optimization execution module 240.

[0109] The model building module 210 can be used to build an evolutionary strategy model for a virtual power plant with the objective functions of maximizing economic benefits, minimizing power deviation, and minimizing operating costs.

[0110] The data acquisition module 220 can be used to collect real-time operating data of each distributed energy source in the virtual power plant;

[0111] The strategy prediction module 230 can be used to take preprocessed real-time operating data and system status as model input of the evolutionary strategy model, use the evolutionary strategy algorithm to train and optimize the scheduling strategy, and output the optimal scheduling strategy for the virtual power plant.

[0112] The optimization execution module 240 can be used to regulate the distributed energy sources in the virtual power plant using scheduling optimization strategies;

[0113] The real-time operating data includes at least one of the following: output power, load demand, energy storage status, and electricity price.

[0114] Optionally, the strategy prediction module 230 can also be used to randomly generate multiple sets of initial strategy parameters to form an initial population;

[0115] The fitness of each initial policy parameter in the initial population is evaluated using an objective function;

[0116] Select initial policy parameters with high fitness for mutation operation, and perform population iteration;

[0117] During the population iteration process, the strategy parameters that meet the fitness criteria are selected as the optimal scheduling strategy for the virtual power plant.

[0118] Optionally, the policy prediction module 230 can also be used to set the initial population size to N and generate N sets of random policy parameters θ at the initial time. i , i = 1, 2, ..., N;

[0119] The initial population formula is expressed as follows:

[0120]

[0121] Optionally, the policy prediction module 230 can also be used to calculate the objective function value of each set of initial policy parameters in the initial population as fitness, specifically including:

[0122] For each time t, based on the policy parameters and the system status of the virtual power plant t Generate the corresponding system action a t ;

[0123] Based on system state s t and system action a t , obtain the execution system action a t The target system state s t+1 ;

[0124] Calculate the objective function value based on the state of the target system.

[0125] Optionally, the strategy prediction module 230 can also be used to perform the following steps S1 to S3:

[0126] Step S1: Select the k strategy parameters with high fitness using either the roulette wheel selection method or the tournament selection method;

[0127] Step S2: Add random noise to the k policy parameters to obtain the next generation of policy parameters, expressed by the following formula:

[0128]

[0129] in, σ represents the next-generation strategy parameters; -(0,I) represents a normal distribution with a mean of 0 and a variance equal to the identity matrix.

[0130] Step S3: Replace the initial population with the new generation of policy parameters to form a new generation population. And repeat steps S1 to S2.

[0131] Optionally, the strategy prediction module 230 can also be used to calculate the fitness of each set of strategy parameters in real time during the population iteration process. If the fitness change is less than a preset threshold, or the maximum number of iterations is reached, the population iteration is terminated.

[0132] After the population iteration terminates, the policy parameter with the highest fitness in the latest population is output as the optimal scheduling policy for the virtual power plant.

[0133] Optionally, the optimization execution module 240 can also be used to generate target system actions based on the optimal scheduling strategy, as expressed in the following formula:

[0134]

[0135] in, Actions for the target system; For policy functions; These are the policy parameters corresponding to the optimal scheduling policy;

[0136] The charging power, discharging power, and / or output power of each distributed energy source in the virtual power plant are regulated by the actions of the target system.

[0137] It should be noted that other corresponding descriptions of the functional modules involved in the virtual power plant dispatch optimization system provided in this embodiment of the invention can be found in [reference needed]. Figure 1 The corresponding descriptions of the methods shown will not be repeated here.

[0138] Based on the above, Figure 1 Accordingly, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the virtual power plant scheduling optimization method of any of the above embodiments.

[0139] Based on the above, Figure 1 The method shown and as Figure 2The embodiment of the system shown in the invention also provides a physical structure diagram of a computer device, such as... Figure 3 As shown, the computer device may include a communication bus, a processor, a memory, and a communication interface. It may also include input / output interfaces and a display device. The various functional units can communicate with each other via the bus. The memory stores a computer program, and the processor executes the program stored in the memory to perform the steps of the virtual power plant scheduling optimization method described in the above embodiments.

[0140] Those skilled in the art will clearly understand that the specific working process of the systems, devices, modules and units described above can be referred to the corresponding process in the foregoing method embodiments. For the sake of brevity, it will not be repeated here.

[0141] Furthermore, the functional units in the various embodiments of the present invention can be physically independent of each other, or two or more functional units can be integrated together, or all functional units can be integrated into one processing unit. The integrated functional units described above can be implemented in hardware, or in software or firmware.

[0142] Those skilled in the art will understand that if the integrated functional unit is implemented in software and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or all or part of it, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computing device (e.g., a personal computer, server, or network device) to execute all or part of the steps of the methods described in the embodiments of the present invention when running the instructions. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0143] Alternatively, all or part of the steps of the foregoing method embodiments can be implemented by hardware (such as a computing device, personal computer, server, or network device) related to program instructions. The program instructions can be stored in a computer-readable storage medium. When the program instructions are executed by the processor of the computing device, the computing device executes all or part of the steps of the methods described in the various embodiments of the present invention.

[0144] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that within the spirit and principles of the present invention, modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein; and these modifications or substitutions do not cause the corresponding technical solutions to depart from the protection scope of the present invention.

Claims

1. A virtual power plant scheduling optimization method, characterized in that, The method includes: For the virtual power plant, an evolutionary strategy model is constructed with the objective functions of maximizing economic benefits, minimizing power deviation, and minimizing operating costs; the objective function formula is expressed as follows: Wherein, revenue t is the revenue calculated based on electricity price and electricity trading volume, in yuan; power deviation t is the deviation between actual power supply and demand, in MW; operating cost t includes energy loss from charging and discharging of energy storage equipment and operation and maintenance costs, in yuan; weighting coefficients α, β, and λ represent the weights of revenue t, power deviation t, and operating cost t, respectively, and are used to adjust the priority of optimization objectives; strategy parameter θ represents the scheduling strategy for the virtual power plant; Collect real-time operating data of each distributed energy source in the virtual power plant; The preprocessed real-time operating data and system status are used as model inputs to the evolutionary strategy model. The evolutionary strategy algorithm is used to train and optimize the scheduling strategy, and the optimal scheduling strategy for the virtual power plant is output, including: Multiple sets of initial policy parameters are randomly generated to form an initial population; Evaluating the fitness of each initial policy parameter group in the initial population using the objective function includes: Calculating the objective function value of each initial policy parameter group in the initial population as the fitness specifically includes: For each time t, based on the policy parameters and the system state s of the virtual power plant t Generate the corresponding system action a t ; Based on the system state s t and the system action a t The system action a is executed. t The target system state s t+1 ; Calculate the objective function value based on the target system state; Select initial policy parameters with high fitness for mutation operation, and perform population iteration; During the population iteration process, the strategy parameters that meet the fitness criteria are selected as the optimal scheduling strategy for the virtual power plant. Wherein, the system state s t The formula is expressed as follows: in, The output power of the distributed energy source at time t, in MW; The load demand at time t, in MW; S t The energy storage state at time t, expressed in MWh; M t The price of electricity at time t is expressed in yuan / MWh. The system action a t The formula is expressed as follows: in, For policy functions; The charging power of the energy storage device at time t, expressed in MW; The discharge power of the energy storage device at time t, expressed in MW; The output power of the distributed energy source at time t is expressed in MW. The target system state s t+1 The formula is expressed as follows: in, The output power of the distributed energy source at time t+1 is expressed in MW. The load demand at time t+1 is expressed in MW; S t+1 The energy storage state at time t+1 is expressed in MWh; M t+1 The electricity price at time t+1 is expressed in yuan / MWh. The scheduling optimization strategy is used to regulate the distributed energy resources in the virtual power plant, including: The target system action is generated based on the optimal scheduling strategy, as expressed by the following formula: in, Actions for the target system; For policy functions; These are the policy parameters corresponding to the optimal scheduling policy; The charging power, discharging power, and / or output power of each distributed energy source in the virtual power plant are regulated by the actions of the target system. The real-time operating data includes at least one of the following: output power, load demand, energy storage status, and electricity price.

2. The method according to claim 1, characterized in that, The random generation of multiple sets of initial policy parameters to form an initial population includes: Set the initial population size to N, and generate N sets of random policy parameters θ at the initial time. i , i = 1, 2, ..., N; The initial population formula is expressed as follows:

3. The method according to claim 1, characterized in that, The step of selecting initial policy parameters with higher fitness for mutation operation and performing population iteration includes: Step S1: Select the k strategy parameters with high fitness using either the roulette wheel selection method or the tournament selection method; Step S2: Add random noise to the k policy parameters to obtain the next generation of policy parameters, expressed by the following formula: in, ε is the parameter of the next-generation strategy; σ is the mutation intensity; ε(0,I) is a normal distribution with a mean of 0 and a variance of the identity matrix; Step S3: Replace the initial population with the new generation strategy parameters to form a new generation population. And repeat steps S1 to S2.

4. The method according to claim 1, characterized in that, The step of selecting policy parameters that meet the fitness criteria as the optimal scheduling strategy for the virtual power plant during the population iteration process includes: During the population iteration process, the fitness of each set of strategy parameters is calculated in real time. If the change in fitness is less than a preset threshold, or the maximum number of iterations is reached, the population iteration is terminated. After the population iteration terminates, the strategy parameter with the highest fitness in the latest population is output as the optimal scheduling strategy for the virtual power plant.

5. A virtual power plant dispatch optimization system, characterized in that, The system includes: The model building module is used to construct an evolutionary strategy model for the virtual power plant, with the objective functions of maximizing economic benefits, minimizing power deviation, and minimizing operating costs; the objective function formula is expressed as follows: Wherein, revenue t is the revenue calculated based on electricity price and electricity trading volume, in yuan; power deviation t is the deviation between actual power supply and demand, in MW; operating cost t includes energy loss from charging and discharging of energy storage equipment and operation and maintenance costs, in yuan; weighting coefficients α, β, and λ represent revenue respectively. t The weights of power deviation t and operating cost t are used to adjust the priority of the optimization objectives; the strategy parameter θ represents the scheduling strategy for the virtual power plant. The data acquisition module is used to collect real-time operating data of each distributed energy source in the virtual power plant; The strategy prediction module is used to take preprocessed real-time operating data and system status as model input to the evolutionary strategy model, train and optimize the scheduling strategy using the evolutionary strategy algorithm, and output the optimal scheduling strategy for the virtual power plant, including: Multiple sets of initial policy parameters are randomly generated to form an initial population; Evaluating the fitness of each initial policy parameter group in the initial population using the objective function includes: Calculating the objective function value of each initial policy parameter group in the initial population as the fitness specifically includes: For each time t, based on the policy parameters and the system state s of the virtual power plant t Generate the corresponding system action a t ; Based on the system state s t and the system action a t The system action a is executed. t The target system state s t+1 ; Calculate the objective function value based on the target system state; Select initial policy parameters with high fitness for mutation operation, and perform population iteration; During the population iteration process, the strategy parameters that meet the fitness criteria are selected as the optimal scheduling strategy for the virtual power plant. Wherein, the system state s t The formula is expressed as follows: in, The output power of the distributed energy source at time t, in MW; The load demand at time t, in MW; S t The energy storage state at time t, expressed in MWh; M t The price of electricity at time t is expressed in yuan / MWh. The system action a t The formula is expressed as follows: in, For policy functions; The charging power of the energy storage device at time t, expressed in MW; The discharge power of the energy storage device at time t, expressed in MW; The output power of the distributed energy source at time t is expressed in MW. The target system state s t+1 The formula is expressed as follows: in, The output power of the distributed energy source at time t+1 is expressed in MW. The load demand at time t+1 is expressed in MW; S t+1 The energy storage state at time t+1 is expressed in MWh; M t+1 The electricity price at time t+1 is expressed in yuan / MWh. The optimization execution module is used to regulate each distributed energy source in the virtual power plant using the scheduling optimization strategy, including: The target system action is generated based on the optimal scheduling strategy, as expressed by the following formula: in, Actions for the target system; For policy functions; These are the policy parameters corresponding to the optimal scheduling policy; The charging power, discharging power, and / or output power of each distributed energy source in the virtual power plant are regulated by the actions of the target system. The real-time operating data includes at least one of the following: output power, load demand, energy storage status, and electricity price.

Images