Virtual power plant resource collaborative scheduling method, system and device and medium

Abstract

The invention discloses a virtual power plant resource collaborative scheduling method, system and device and a medium, and the method comprises the steps: obtaining historical load data of a virtual power plant, and predicting a load prediction value of the virtual power plant in a preset time period based on the historical load data; inputting the load prediction value into a preset collaborative scheduling model, and performing optimization by taking operation cost minimization as a target to obtain an initial power value corresponding to each energy unit; and based on the initial power value, determining a feasible region range of the virtual power plant through a reinforcement learning method during each iteration, inputting the feasible region range into the collaborative scheduling model for iterative optimization until an iteration termination condition is met, determining a target power value corresponding to each energy unit, and outputting the target power value to the virtual power plant. And scheduling each energy unit based on the target power value. The flexibility of virtual power plant resource scheduling can be improved, and the resource utilization efficiency is improved.

Description

Technical Field

[0001] The present application relates to the field of virtual power plants, and particularly to a method, system, device and medium for collaborative scheduling of virtual power plant resources. Background Art

[0002] As a new energy management paradigm, virtual power plants have become a core technology for optimizing the dispatching of power systems and enhancing system flexibility. It can integrate numerous dispersed energy units, such as wind energy, solar energy, energy storage devices, etc., and through an intelligent communication network and advanced control algorithms, achieve information sharing and collaborative operation. By optimizing the feasible region of the virtual power plant, it can flexibly adjust the output power of each energy unit according to real-time power demands and market dynamics, thereby optimizing the energy dispatching operation plan of the entire system. This optimization not only improves the energy utilization efficiency, realizes the rational allocation and full utilization of energy, but also enhances the stability and reliability of the power system, reducing system problems caused by fluctuations or failures of a single energy source.

[0003] However, most of the current feasible region evaluation methods for virtual power plants rely on traditional rule-based or model-based approaches. These methods often require manual setting of many parameters, and are overwhelmed and difficult to adapt when facing complex and changeable power market environments and various energy units.

[0004] Application Content

[0005] The present application provides a method, system, device and medium for collaborative scheduling of virtual power plant resources to improve the flexibility of virtual power plant resource scheduling and the resource utilization efficiency.

[0006] In a first aspect, the present application provides a method for collaborative scheduling of virtual power plant resources, including:

[0007] Obtain the historical load data of the virtual power plant, and predict the load prediction value of the virtual power plant in a preset time period based on the historical load data;

[0008] Input the load prediction value into a preset collaborative scheduling model, and optimize with the goal of minimizing the operating cost to obtain the initial power values corresponding to each energy unit, where the initial power values include the virtual power plant interaction power, photovoltaic power generation power, wind turbine power generation power, and energy storage system charge and discharge power;

[0009] Based on the initial power value, at each iteration, the feasible region range of the virtual power plant is determined by the reinforcement learning method, and the feasible region range is input into the coordinated scheduling model for iterative optimization until the iteration termination condition is met, and the target power value corresponding to each energy unit is determined, and each energy unit is scheduled based on the target power value, wherein the feasible region range includes the power upper limit value and the power lower limit value corresponding to each energy unit.

[0010] In the embodiment of the present application, by predicting the load prediction value of the virtual power plant in a preset time period, the change trend of the load demand of the virtual power plant in the preset time period can be understood in advance, so that the scheduling strategy can be adjusted and optimized in advance according to the prediction result, and different load scenarios can be flexibly responded to; by optimizing with the goal of minimizing the operation cost, on the premise of meeting the load demand, the characteristics and cost factors of each energy unit can be comprehensively considered, and the advantages of each energy unit can be fully utilized to flexibly allocate the power output of each energy unit; by dynamically adjusting the feasible region range of the virtual power plant, the scheduling strategy of the virtual power plant can flexibly respond to various uncertainties and changes, enhance the adaptability of the system to complex environments, and at the same time can more accurately limit the power output range of each energy unit to avoid exceeding its own operation ability and economic operation range, thereby improving the resource utilization rate. Compared with the prior art, the present application can improve the flexibility of virtual power plant resource scheduling and improve the resource utilization efficiency.

[0011] Further, before inputting the load prediction value into a preset coordinated scheduling model, it further includes:

[0012] Obtain the time-of-use price of the external power grid, the equipment operation cost, and the upper and lower limits of equipment operation respectively, where the upper and lower limits of equipment operation include the upper and lower limits of photovoltaic power, the upper and lower limits of wind turbine power, the upper and lower limits of energy storage power, and the upper and lower limits of energy storage capacity;

[0013] According to the load prediction value, the time-of-use price of the external power grid, the equipment operation cost, and the upper and lower limits of equipment operation, establish an objective function and constraint conditions with the goal of minimizing the operation cost, where the constraint conditions include the virtual power plant interaction power constraint, the photovoltaic power generation constraint, the wind turbine power generation constraint, the energy storage system charge and discharge power constraint, and the energy storage system capacity constraint.

[0014] In this way, according to the load prediction value, the time-of-use price of the external power grid, the equipment operation cost, and the upper and lower limits of equipment operation, establishing an objective function and constraint conditions with the goal of minimizing the operation cost can obtain an accurate coordinated scheduling model, which is convenient for subsequent comprehensive consideration of the characteristics and cost factors of each energy unit and fully utilizing the advantages of each energy unit to flexibly allocate the power output of each energy unit.

[0015] Further, the expression of the coordinated scheduling model is specifically:

[0016]

[0017] Wherein, λ TOU is the time-of-use price of the external power grid; is the interactive power of the virtual power plant at time t, representing the total power exchanged in real time between the virtual power plant and the power grid; C PV , C WT and C BS are respectively the operating cost of the photovoltaic system, the operating cost of the wind turbine, and the operating cost of the energy storage system; C DR is the cost of the load participating in demand response; and are respectively the photovoltaic power generation power and the wind turbine power generation power at time t; and are respectively the charging power and the discharging power of the energy storage system at time t; θ t is the load regulation ratio parameter at time t; is the load prediction value at time t; and are the state of charge of the energy storage system at times t+1 and t; α BS is the self-discharge coefficient of the energy storage system; η BS is the charge-discharge efficiency of the energy storage system; and are the upper and lower limits of the virtual power plant interactive power constraint; and are the upper and lower limits of the energy storage capacity; and are the upper and lower limits of the energy storage system charging power; and are the upper and lower limits of the energy storage system discharging power; and are the upper and lower limits of the photovoltaic power; and are the upper and lower limits of the wind turbine power.

[0018] Furthermore, the feasible region range of the virtual power plant is determined by the reinforcement learning method, specifically:

[0019] Construct a state space of reinforcement learning based on the virtual power plant interactive power and the preset load regulation ratio parameter;

[0020] Based on the state space, the preset action space, and the preset reward function, use the preset reinforcement learning model to perform iterative optimization with the goal of maximizing the expected value of the cumulative value of the reward function to determine the feasible region range of the virtual power plant.

[0021] By dynamically adjusting the feasible region of the virtual power plant in this way, the scheduling strategy of the virtual power plant can flexibly respond to various uncertainties and changes, enhancing the adaptability of the system to complex environments. At the same time, it can more accurately limit the power output range of each energy unit, preventing it from exceeding its own operating capacity and economic operating range, thereby improving resource utilization efficiency.

[0022] Further, using the preset reinforcement learning model to iteratively optimize with the goal of maximizing the expected value of the cumulative reward function to determine the feasible region of the virtual power plant specifically includes:

[0023] Using the maximum entropy inverse reinforcement learning algorithm to solve the reward function to obtain the state-action value function, and determining the policy gradient based on the state-action value function;

[0024] Based on the policy gradient, update the reward function through the gradient ascent method, and solve the updated reward function to determine the feasible region of the virtual power plant.

[0025] By dynamically adjusting the feasible region of the virtual power plant in this way, the scheduling strategy of the virtual power plant can flexibly respond to various uncertainties and changes, enhancing the adaptability of the system to complex environments. At the same time, it can more accurately limit the power output range of each energy unit, preventing it from exceeding its own operating capacity and economic operating range, thereby improving resource utilization efficiency.

[0026] Further, the relevant formulas for determining the feasible region of the virtual power plant by the reinforcement learning method are specifically:

[0027]

[0028] a t =π(s t );

[0029] max R(s,a)=η T φ(s,a)+∈;

[0030]

[0031] In the formula, s t is the state space; is the interactive power of the virtual power plant at time t; θ t is the load regulation ratio parameter at time t; a t is the action space; π is the policy of reinforcement learning, used to select the optimal action a t according to the current state s t ; R(s,a) is the reward function; η Tis the reward function parameter; φ(s,a) is the feature vector; ∈ is the noise, used to simulate the randomness in the reward function; and are respectively the minimum and maximum values of the virtual power plant's interaction power; and are respectively the minimum and maximum values of the virtual battery's interaction power in the historical data; is the said policy gradient, representing the policy parameter η μ the gradient of the objective function J; Q(s,a∣η Q ) is the said state-action value function, used to evaluate the expected cumulative reward of performing action a in state s; μ(s∣η μ ) is the policy parameterization, representing the probability distribution of selecting an action in state s according to the reward function parameter η μ ; is the gradient of the action space.

[0032] Further, predicting the load prediction value of the virtual power plant in a preset time period based on the historical load data specifically includes:

[0033] Performing data preprocessing on the historical load data to obtain processed data;

[0034] Inputting the processed data into a preset prediction model, and using the model parameters for prediction to obtain the load prediction value of the virtual power plant in the preset time period.

[0035] In this way, by predicting the load prediction value of the virtual power plant in the preset time period, the change trend of the load demand of the virtual power plant in the preset time period can be understood in advance, so that the scheduling strategy can be adjusted and optimized in advance according to the prediction results, and flexibly respond to different load scenarios.

[0036] In a second aspect, the present application provides a virtual power plant resource collaborative scheduling system, including: an acquisition module, a processing module, and a scheduling module;

[0037] The acquisition module is used to acquire the historical load data of the virtual power plant and predict the load prediction value of the virtual power plant in a preset time period based on the historical load data;

[0038] The processing module is used to input the load prediction value into a preset collaborative scheduling model, optimize with the goal of minimizing the operating cost, and obtain the initial power values corresponding to each energy unit, where the initial power values include the virtual power plant's interaction power, photovoltaic power generation power, wind turbine power generation power, and energy storage system charge and discharge power;

[0039] The scheduling module is used to, based on the initial power value, determine the feasible region range of the virtual power plant by means of reinforcement learning at each iteration, and input the feasible region range into the collaborative scheduling model for iterative optimization until the iterative termination condition is met, determine the target power values corresponding to each energy unit, and schedule each energy unit based on the target power values, where the feasible region range includes the power upper limit value and the power lower limit value corresponding to each energy unit.

[0040] In the embodiments of the present application, by predicting the load prediction value of the virtual power plant in a preset time period, the change trend of the load demand of the virtual power plant in the preset time period can be understood in advance, so that the scheduling strategy can be adjusted and optimized in advance according to the prediction result, and different load scenarios can be flexibly responded to; by optimizing with the goal of minimizing the operating cost, on the premise of meeting the load demand, the characteristics and cost factors of each energy unit can be comprehensively considered, and the advantages of each energy unit can be fully utilized to flexibly allocate the power output of each energy unit; by dynamically adjusting the feasible region range of the virtual power plant, the scheduling strategy of the virtual power plant can flexibly respond to various uncertainties and changes, enhance the adaptability of the system to complex environments, and at the same time can more accurately limit the power output range of each energy unit to avoid exceeding its own operating capacity and economic operating range, thereby improving the resource utilization rate. Compared with the prior art, the present application can improve the flexibility of virtual power plant resource scheduling and the resource utilization efficiency.

[0041] In a third aspect, the present application also provides a terminal device, including: one or more processors; a memory coupled to the processor for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the virtual power plant resource collaborative scheduling method as described in the present application.

[0042] In a fourth aspect, the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the virtual power plant resource collaborative scheduling method as described in the present application is implemented. Description of the Drawings

[0043] Figure 1 is a schematic structural diagram of a virtual power plant provided by the present application;

[0044] Figure 2 is a schematic flow diagram of an embodiment of the virtual power plant resource collaborative scheduling method provided by the present application;

[0045] Figure 3 is a schematic structural diagram of an embodiment of the virtual power plant resource collaborative scheduling system provided by the present application;

[0046] Figure 4It is a schematic structural diagram of an embodiment of the terminal device provided by this application. Detailed implementation manners

[0047] Next, the technical solutions in the embodiments of this application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part rather than all of the embodiments of this application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts shall fall within the protection scope of this application.

[0048] It should be understood that the step numbers used in the text are only for convenience of description and do not limit the order of execution of the steps.

[0049] It should be understood that the terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0050] The terms "comprising" and "including" indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their combinations.

[0051] The term " / or" refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0052] A virtual power plant is a core technology for optimizing the dispatching of power systems and enhancing flexibility. By integrating dispersed energy sources such as wind energy, solar energy, and energy storage devices, and relying on intelligent communication networks and advanced control algorithms, it realizes information sharing and coordinated operation. By optimizing the feasible region of the virtual power plant, the output power of each energy unit can be flexibly adjusted according to real-time power demands and market dynamics, and the system energy dispatching plan can be optimized. However, the current feasible region evaluation methods for virtual power plants mostly rely on traditional rules or models, require manual parameter setting, and have insufficient adaptability in the face of complex power markets and diverse energy units.

[0053] Next, the nouns involved in this application are analyzed:

[0054] A virtual power plant is an innovative energy management paradigm that integrates distributed generation resources (such as wind energy, solar energy, energy storage devices, etc.) into a virtual entity that can be uniformly dispatched and managed.

[0055] The feasible region of a virtual power plant refers to the range of power output that a virtual power plant can achieve under given operating constraints and market conditions. This range takes into account the operating constraints of each distributed energy unit within the virtual power plant (such as energy storage systems, photovoltaic systems, wind turbines, etc.), as well as the demands of the external power grid and market dynamics.

[0056] Based on this, the embodiments of the present application provide a method, system, device, and medium for collaborative scheduling of virtual power plant resources, which can improve the flexibility of virtual power plant resource scheduling and the resource utilization efficiency.

[0057] The method, system, device, and medium for collaborative scheduling of virtual power plant resources provided by the embodiments of the present application are specifically described through the following embodiments. First, the method for collaborative scheduling of virtual power plant resources in the embodiments of the present application is described.

[0058] The method for collaborative scheduling of virtual power plant resources provided by the embodiments of the present application relates to the field of virtual power plants. The method for collaborative scheduling of virtual power plant resources provided by the embodiments of the present application can be applied to terminals, can also be applied to server sides, or can also be software running on terminals or server sides. In some embodiments, the terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, etc.; the server side can be configured as an independent physical server, can also be configured as a server cluster or distributed system composed of multiple physical servers, or can also be configured as a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the method for collaborative scheduling of virtual power plant resources, etc., but is not limited to the above forms.

[0059] The present application can also be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multi-processor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and so on. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0060] The structural schematic diagram of the virtual power plant is as Figure 1As shown in the figure, a virtual power plant integrates various distributed energy sources such as wind energy, solar energy, and energy storage devices into a unified energy supply system through an intelligent control system. It can flexibly allocate the output power of wind energy, solar energy, and energy storage devices according to real-time power demand and the power generation capacity of each energy unit, enabling them to work together to provide a stable power supply.

[0061] Embodiment 1

[0062] Please refer to Figure 2 , Figure 2 which is a schematic flowchart of an embodiment of the virtual power plant resource collaborative scheduling method provided by this application, including steps S101 to S103;

[0063] Step S101: Obtain the historical load data of the virtual power plant, and predict the load prediction value of the virtual power plant in a preset time period based on the historical load data;

[0064] In some embodiments, obtaining the historical load data of the virtual power plant specifically means: obtaining the historical load data of different time periods (such as hours, days, etc.) from the intelligent electricity meters, SCADA systems, enterprise energy management systems, etc. of the virtual power plant, and at the same time collecting data related to the load, such as meteorological data (temperature, humidity, wind speed, etc.), calendar information (working days, holidays, etc.), and social activity data (major events, policy changes, etc.).

[0065] In some embodiments, predicting the load prediction value of the virtual power plant in a preset time period based on the historical load data includes: performing data preprocessing on the historical load data to obtain processed data; inputting the processed data into a preset prediction model, and using the model parameters for prediction to obtain the load prediction value of the virtual power plant in the preset time period. Specifically, first, clean the collected historical load data to remove outliers and correct incorrect data, and then perform data normalization on the cleaned historical load data to obtain processed data; second, fuse the processed data with meteorological data, calendar information, and social activity data to construct a data set that can be directly input into the model for training; then, use the data set to train the selected prediction model. During the training process, adjust the parameters of the prediction model (such as learning rate, the number of neurons in the hidden layer, etc.) to enable the prediction model to accurately fit the historical data and minimize the prediction error, thereby obtaining a trained prediction model; finally, input the processed data into the trained prediction model to predict the load demand of the virtual power plant in a preset time period (such as the next few hours or days) based on the historical load data and relevant influencing factor data.

[0066] It should be noted that the prediction model can be but is not limited to time series analysis, linear regression, neural network, long short-term memory network (LSTM), etc.

[0067] By predicting the load forecast value of the virtual power plant in the preset time period, the change trend of the load demand of the virtual power plant in the preset time period can be understood in advance, so that the dispatching strategy can be adjusted and optimized in advance according to the prediction result, and different load scenarios can be flexibly responded to.

[0068] Step S102: Input the load forecast value into a preset collaborative dispatching model, optimize it with the goal of minimizing the operating cost, and obtain the initial power values corresponding to each energy unit, where the initial power values include the interactive power of the virtual power plant, the photovoltaic power generation, the wind turbine power generation, and the charge and discharge power of the energy storage system.

[0069] In some embodiments, before inputting the load forecast value into the preset collaborative dispatching model, it further includes: respectively obtaining the time-of-use price of the external power grid, the equipment operating cost, and the upper and lower limits of equipment operation, where the upper and lower limits of equipment operation include the upper and lower limits of photovoltaic power, the upper and lower limits of wind turbine power, the upper and lower limits of energy storage power, and the upper and lower limits of energy storage capacity; according to the load forecast value, the time-of-use price of the external power grid, the equipment operating cost, and the upper and lower limits of equipment operation, establish an objective function and constraint conditions with the goal of minimizing the operating cost, where the constraint conditions include the interactive power constraint of the virtual power plant, the photovoltaic power generation constraint, the wind turbine power generation constraint, the charge and discharge power constraint of the energy storage system, and the energy storage system capacity constraint.

[0070] It should be noted that the energy units include an energy storage system, a photovoltaic system, and a wind turbine.

[0071] In this way, according to the load forecast value, the time-of-use price of the external power grid, the equipment operating cost, and the upper and lower limits of equipment operation, establishing an objective function and constraint conditions with the goal of minimizing the operating cost can obtain an accurate collaborative dispatching model, which is convenient for subsequent comprehensive consideration of the characteristics and cost factors of each energy unit, and flexibly allocating the power output of each energy unit by making full use of the advantages of each energy unit.

[0072] In some embodiments, the expression of the collaborative dispatching model is specifically:

[0073]

[0074]

[0075] In the formula, λ TOU is the time-of-use price of the external power grid; is the interactive power of the virtual power plant at time t, indicating the total power exchanged in real time between the virtual power plant and the power grid; C PV 、C WT and C BSare the operating costs of the photovoltaic system, the wind turbine system, and the energy storage system respectively; C DR is the cost of the load participating in demand response; and are the photovoltaic power generation power and the wind turbine power generation power at time t respectively; and are the charging power and the discharging power of the energy storage system at time t respectively; θ t is the load regulation ratio parameter at time t; is the load prediction value at time t; and are the state of charge of the energy storage system at time t + 1 and time t respectively; α BS is the self-discharge coefficient of the energy storage system; η BS is the charge-discharge efficiency of the energy storage system; and are the upper and lower limits of the virtual power plant interaction power constraint; and are the upper and lower limits of the energy storage capacity; and are the upper and lower limits of the energy storage system charging power; and are the upper and lower limits of the energy storage system discharging power; and are the upper and lower limits of the photovoltaic power; and are the upper and lower limits of the wind turbine power.

[0076] It should be noted that after determining the coordinated scheduling model, the coordinated scheduling model can be solved to obtain the initial power values corresponding to each energy unit (energy storage system, photovoltaic system, wind turbine, etc.). Among them, the initial power values include: virtual power plant interaction power photovoltaic power generation power wind turbine power generation power energy storage system charging power and energy storage system discharging power

[0077] It should be noted that the time-of-use price λ of the external power grid TOU is a price list formulated by the power company according to the electricity consumption demand and cost in different periods, and can usually be obtained from the official documents or websites of the power company. The virtual power plant interaction power p t g , represents the power of the virtual power plant interacting with the external power grid at time t. A positive value indicates purchasing electricity from the grid, and a negative value indicates selling electricity to the grid. The operating cost C of the photovoltaic system PV and the operating cost C of the wind turbine WT, including costs such as equipment maintenance, cleaning, and fault repair, which are usually estimated based on the scale of the PV system / wind turbine, the maintenance plan provided by the equipment supplier, and historical maintenance records. The operating cost C of the energy storage system BS including costs such as charge and discharge losses, cooling, and maintenance of the battery, which are calculated based on the specifications of the energy storage system, the technical parameters provided by the battery supplier, and the maintenance manual. The cost C of the load participating in demand response DR , which involves the additional costs incurred by users for adjusting their electricity consumption plans in response to grid demands, such as compensation costs and equipment adjustment costs, and is usually determined according to the specific rules of the demand response project and the negotiation agreement with users. The load regulation ratio parameter θ t , which represents the proportion of the user's load that can be adjusted under the demand response mechanism and is set according to the rules of the demand response project and the user participation survey or historical response data. The self-discharge coefficient α of the energy storage system BS describes the degree of energy loss generated by the energy storage device due to its own characteristics during the energy storage process, which is usually a constant slightly less than 1 and is provided by the battery supplier or determined through experiments. The charge and discharge efficiency η of the energy storage system BS reflects the energy conversion efficiency of the energy storage device during the charging and discharging processes, with a value between 0 and 1, which is provided by the battery supplier or determined through experiments. The upper and lower limits of the virtual power plant interaction power constraint, the upper and lower limits of the energy storage capacity, the upper and lower limits of the energy storage system charging power, the upper and lower limits of the energy storage system discharging power, the upper and lower limits of the PV power, and the upper and lower limits of the wind turbine power can all be obtained according to the operation manual and technical documents of the corresponding energy storage unit.

[0078] Step S103, based on the initial power value, at each iteration, determine the feasible region range of the virtual power plant through the reinforcement learning method, and input the feasible region range into the coordinated scheduling model for iterative optimization until the iteration termination condition is met, determine the target power values corresponding to each energy unit, and schedule each energy unit based on the target power values, where the feasible region range includes the power upper limit value and the power lower limit value corresponding to each energy unit.

[0079] In some embodiments, the feasible region range includes the power upper limit value and the power lower limit value corresponding to each energy unit, that is, the upper and lower limits of the virtual power plant interaction power constraint, the upper and lower limits of the energy storage capacity, the upper and lower limits of the energy storage system charging power, the upper and lower limits of the energy storage system discharging power, the upper and lower limits of the PV power, and the upper and lower limits of the wind turbine power mentioned above. Since the distributed energy sources (such as PV and wind power) aggregated by the virtual power plant are affected by real-time factors such as weather, sunlight, and wind speed, their output powers are intermittent. Therefore, it is necessary to update the feasible region range to ensure stable response to grid demands during energy fluctuations.

[0080] In some embodiments, determining the feasible region range of the virtual power plant through a reinforcement learning method includes: constructing a state space for reinforcement learning based on the interactive power of the virtual power plant and a preset load regulation ratio parameter; based on the state space, a preset action space, and a preset reward function, using a preset reinforcement learning model to perform iterative optimization with the goal of maximizing the expected value of the cumulative value of the reward function to determine the feasible region range of the virtual power plant. Specifically, based on the interactive power of the virtual power plant and a preset load regulation ratio parameter θ t Construct a state space for reinforcement learning, where the state space is: After that, define the action space a t = π(s t ), where π is the policy of reinforcement learning, and the action space includes a first output power value (including the interactive power of the virtual power plant photovoltaic power generation wind turbine power generation energy storage system charging power and energy storage system discharging power ), and set the reward function, max R(s,a) = η T φ(s,a)+∈, where R(s,a) is the reward function; η T is the reward function parameter; φ(s,a) is the feature vector; ∈ is the noise. After that, use a preset reinforcement learning model to perform iterative optimization with the goal of maximizing the expected value of the cumulative value of the reward function to determine the feasible region range of the virtual power plant.

[0081] Furthermore, the iterative optimization using a preset reinforcement learning model with the goal of maximizing the expected value of the cumulative value of the reward function to determine the feasible region range of the virtual power plant includes: using the maximum entropy inverse reinforcement learning algorithm to solve the reward function to obtain the state-action value function, and determining the policy gradient based on the state-action value function; based on the policy gradient, updating the reward function through the gradient ascent method, and solving the updated reward function to determine the feasible region range of the virtual power plant. Specifically, first, when determining the action space a t and the state space s t After that, the historical state-action trajectory set D = {(s t ,a t ,s t+1 )+ can be constructed. After that, by maximizing the likelihood function or the gradient ascent method, optimize the reward function parameter η T, so that the expert strategy can maximize the cumulative reward to obtain the optimized reward function R(s,a); secondly, based on the optimized reward function R(s,a), methods such as dynamic programming, Monte Carlo method or temporal difference learning are used to calculate the state-action value function Q(s,a|η Q ), and calculate the policy gradient according to the trajectory set D and the state-action value function Q(s,a|η Q ) Then, when the policy gradient is determined After that, use the gradient ascent formula Update the reward function parameter η by the gradient ascent method T , so that the policy in the historical data can maximize the cumulative reward; finally, according to the updated reward function parameter η T , adjust the policy parameter η μ , so that the policy can better adapt to the new reward function and generate a better action sequence. Among them, in each iteration process, according to the updated reward function parameter η T and the policy parameter η μ , evaluate the upper and lower limits of the virtual power plant's power. By minimizing make the interactive power upper and lower limits closer to the upper and lower limits in the historical data, so as to more accurately approximate the feasible region of the virtual power plant. When the change amount of the power upper and lower limits is less than the preset threshold in continuous multiple iterations, it is considered that the algorithm converges, and the final feasible region range of the virtual power plant is obtained.

[0082] Furthermore, the relevant formulas for determining the feasible region range of the virtual power plant by the reinforcement learning method are specifically as follows:

[0083]

[0084] a t =π(s t );

[0085] max R(s,a)=η T φ(s,a)+∈;

[0086]

[0087] In the formula, s t is the state space; is the interactive power of the virtual power plant at time t; θ t is the load regulation ratio parameter at time t; a t is the action space; π is the policy of reinforcement learning, which is used to select the optimal action a t according to the current state s t ; R(s,a) is the reward function; η Tis the reward function parameter; φ(s,a) is the feature vector; ∈ is the noise, which is used to simulate the randomness in the reward function; and are respectively the minimum and maximum values of the virtual power plant's interaction power; and are respectively the minimum and maximum values of the virtual battery's interaction power in the historical data; is the said policy gradient, representing the policy parameter η μ The gradient of the objective function J; Q(s,a∣η Q ) is the said state-action value function, which is used to evaluate the expected cumulative reward of executing action a in state s; μ(s∣η μ ) is the policy parameterization, representing the probability distribution of selecting an action according to the reward function parameter η μ in state s; is the gradient of the action space.

[0088] In this way, by dynamically adjusting the feasible region range of the virtual power plant, the dispatching strategy of the virtual power plant can flexibly cope with various uncertainties and changes, enhancing the system's adaptability to complex environments. At the same time, it can more accurately limit the power output range of each energy unit, preventing it from exceeding its own operating capacity and economic operating range, thereby improving resource utilization efficiency.

[0089] In the embodiment of the present application, by predicting the load forecast value of the virtual power plant in the preset time period, the change trend of the load demand of the virtual power plant in the preset time period can be understood in advance, enabling the dispatching strategy to be adjusted and optimized in advance according to the prediction results and flexibly coping with different load scenarios; by optimizing with the goal of minimizing the operating cost, under the premise of meeting the load demand, the characteristics and cost factors of each energy unit can be comprehensively considered, and the advantages of each energy unit can be fully utilized to flexibly allocate the power output of each energy unit; by dynamically adjusting the feasible region range of the virtual power plant, the dispatching strategy of the virtual power plant can flexibly cope with various uncertainties and changes, enhancing the system's adaptability to complex environments. At the same time, it can more accurately limit the power output range of each energy unit, preventing it from exceeding its own operating capacity and economic operating range, thereby improving resource utilization efficiency. Compared with the prior art, the present application can improve the flexibility of virtual power plant resource scheduling and improve resource utilization efficiency.

[0090] Embodiment Two

[0091] Please refer to Figure 3 , Figure 3 which is a schematic structural diagram of an embodiment of the virtual power plant resource collaborative dispatching system provided by the present application, including: an acquisition module 100, a processing module 200, and a dispatching module 300;

[0092] The obtaining module 100 is configured to obtain historical load data of a virtual power plant, and predict a load prediction value of the virtual power plant in a preset time period based on the historical load data;

[0093] The processing module 200 is configured to input the load prediction value into a preset collaborative scheduling model, optimize it with the goal of minimizing the operating cost, and obtain initial power values corresponding to each energy unit, where the initial power values include the interactive power of the virtual power plant, the photovoltaic power generation, the wind turbine power generation, and the charge and discharge power of the energy storage system;

[0094] The scheduling module 300 is configured to, based on the initial power values, determine the feasible region range of the virtual power plant through a reinforcement learning method at each iteration, and input the feasible region range into the collaborative scheduling model for iterative optimization until the iteration termination condition is met, determine the target power values corresponding to each energy unit, and schedule each energy unit based on the target power values, where the feasible region range includes the upper power limit value and the lower power limit value corresponding to each energy unit.

[0095] Regarding the information interaction, execution process, etc. among the modules in the above virtual power plant resource collaborative scheduling system, since they are based on the same concept as the embodiments of the virtual power plant resource collaborative scheduling method in the first aspect of the present invention, the achieved technical effects are basically the same. For specific content, reference can be made to the description in Embodiment 1 of the method of the present invention, and details will not be elaborated here.

[0096] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separated, that is, they may be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the method in this embodiment.

[0097] Please refer to Figure 4 , Figure 4 which shows the hardware structure of a terminal device in another embodiment. The terminal device includes:

[0098] A processor 401, which can be implemented in ways such as a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is configured to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0099] The memory 402 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM), etc. The memory 402 can store an operating system and other application programs. When implementing the technical solutions provided in the embodiments of this specification through software or firmware, the relevant program codes are stored in the memory 402, and the processor 401 is used to call and execute the large model-based dialogue risk assessment method of the embodiments of this application;

[0100] The input / output interface 403 is used to implement information input and output;

[0101] The communication interface 404 is used to implement communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or through wireless means (such as mobile network, WIFI, Bluetooth, etc.);

[0102] The bus 405 transmits information between various components of the device (such as the processor 401, the memory 402, the input / output interface 403, and the communication interface 404);

[0103] Among them, the processor 401, the memory 402, the input / output interface 403, and the communication interface 404 achieve communication connections with each other inside the device through the bus 405.

[0104] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the virtual power plant resource collaborative scheduling method as described in Embodiment 1 above.

[0105] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, the storage medium can be a magnetic disk, an optical disc, a read-only memory (ROM), or a random access memory (RAM), etc.

[0106] The specific embodiments described above have further detailed the purpose, technical solutions, and beneficial effects of this application. It should be understood that the above descriptions are only specific embodiments of this application and are not used to limit the protection scope of this application.

[0107] It is particularly pointed out that, for those skilled in the art, any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of this application shall be included within the protection scope of this application.

Claims

1. A virtual power plant resource collaborative scheduling method, characterized in that, Including: Obtain the historical load data of the virtual power plant, and predict the load prediction value of the virtual power plant in a preset time period based on the historical load data; Input the load prediction value into a preset collaborative scheduling model, optimize it with the goal of minimizing the operating cost, and obtain the initial power values corresponding to each energy unit, where the initial power values include the virtual power plant interaction power, photovoltaic power generation power, wind turbine power generation power, and energy storage system charge and discharge power; Based on the initial power values, at each iteration, determine the feasible region range of the virtual power plant through the reinforcement learning method, and input the feasible region range into the collaborative scheduling model for iterative optimization until the iteration termination condition is met, determine the target power values corresponding to each energy unit, and schedule each energy unit based on the target power values, where the feasible region range includes the power upper limit value and power lower limit value corresponding to each energy unit.

2. The virtual power plant resource collaborative scheduling method according to claim 1, wherein Before inputting the load prediction value into the preset collaborative scheduling model, it further includes: Obtain the time-of-use price of the external power grid, equipment operating costs, and equipment operating upper and lower limits respectively, where the equipment operating upper and lower limits include photovoltaic power upper and lower limits, wind turbine power upper and lower limits, energy storage power upper and lower limits, and energy storage capacity upper and lower limits; According to the load prediction value, the time-of-use price of the external power grid, the equipment operating costs, and the equipment operating upper and lower limits, establish an objective function and constraint conditions for minimizing the operating cost, where the constraint conditions include virtual power plant interaction power constraints, photovoltaic power generation power constraints, wind turbine power generation constraints, energy storage system charge and discharge power constraints, and energy storage system capacity constraints.

3. The virtual power plant resource collaborative scheduling method according to claim 2, wherein The expression of the collaborative scheduling model is specifically: where λ TOU is the time-of-use price of the external power grid; is the interactive power of the virtual power plant at time t, representing the total power exchanged in real time between the virtual power plant and the power grid; C PV , C WT and C BS are the operating costs of the photovoltaic system, the wind turbine system, and the energy storage system, respectively; C DR is the cost of the load participating in demand response; and are the photovoltaic power generation power and the wind turbine power generation power at time t, respectively; and are the charging power and the discharging power of the energy storage system at time t, respectively; θ t is the load regulation ratio parameter at time t; is the load prediction value at time t; and are the state of charge of the energy storage system at times t + 1 and t, respectively; α BS is the self-discharge coefficient of the energy storage system; η BS is the charge-discharge efficiency of the energy storage system; and are the upper and lower limits of the virtual power plant interactive power constraint; and are the upper and lower limits of the energy storage capacity; and are the upper and lower limits of the energy storage system charging power; and are the upper and lower limits of the energy storage system discharging power; and are the upper and lower limits of the photovoltaic power; and are the upper and lower limits of the wind turbine power.

4. The virtual power plant resource collaborative scheduling method according to claim 1, wherein The method for determining the feasible region range of the virtual power plant through the reinforcement learning method is specifically: Construct the state space of reinforcement learning based on the virtual power plant interaction power and a preset load regulation ratio parameter; Based on the state space, a preset action space, and a preset reward function, use a preset reinforcement learning model to perform iterative optimization with the goal of maximizing the expected value of the cumulative value of the reward function, and determine the feasible region range of the virtual power plant.

5. The virtual power plant resource collaborative scheduling method according to claim 4, characterized in that The method for using the preset reinforcement learning model to perform iterative optimization with the goal of maximizing the expected value of the cumulative value of the reward function and determining the feasible region range of the virtual power plant is specifically: Use the maximum entropy inverse reinforcement learning algorithm to solve the reward function to obtain the state-action value function, and determine the policy gradient based on the state-action value function; Based on the policy gradient, update the reward function through the gradient ascent method, and solve the updated reward function to determine the feasible region range of the virtual power plant.

6. The virtual power plant resource collaborative scheduling method according to claim 5, characterized in that, The relevant formulas for determining the feasible region range of the virtual power plant through the reinforcement learning method are specifically: a t = π(s t ); max R(s,a)=η T φ(s,a)+∈; where s t is the state space; is the virtual power plant interaction power at time t; θ t is the load regulation ratio parameter at time t; a t is the action space; π is the policy of reinforcement learning, which is used to select the optimal action a according to the current state s t ; R(s, a) is the reward function; η t is the reward function parameter; φ(s, a) is the feature vector; ∈ is the noise, which is used to simulate the randomness in the reward function; T and are the minimum and maximum values of the virtual power plant interaction power respectively; and are the minimum and maximum values of the virtual battery interaction power in historical data respectively; is the policy gradient, which represents the gradient of the policy parameter η μ with respect to the objective function J; Q(s, a|η Q ) is the state-action value function, which is used to evaluate the expected cumulative reward for executing action a in state s; μ(s|η μ ) is the policy parameterization, which represents the probability distribution of selecting actions according to the reward function parameter η μ in state s; is the gradient of the action space.​ 7. The virtual power plant resource collaborative scheduling method according to claim 1, wherein The method for predicting the load prediction value of the virtual power plant in a preset time period based on the historical load data is specifically: Perform data preprocessing on the historical load data to obtain processed data; Input the processed data into a preset prediction model, and use the model parameters for prediction to obtain the load prediction value of the virtual power plant in the preset time period.

8. A virtual power plant resource collaborative scheduling system, characterized in that Including: An acquisition module, a processing module, and a scheduling module; The obtaining module is configured to obtain the historical load data of the virtual power plant and predict the load prediction value of the virtual power plant in a preset time period based on the historical load data; The processing module is configured to input the load prediction value into a preset collaborative scheduling model, optimize it with the goal of minimizing the operation cost, and obtain the initial power values corresponding to each energy unit, where the initial power values include the interactive power of the virtual power plant, the photovoltaic power generation, the wind turbine power generation, and the charge and discharge power of the energy storage system; The scheduling module is configured to, based on the initial power values, determine the feasible region range of the virtual power plant through a reinforcement learning method at each iteration, input the feasible region range into the collaborative scheduling model for iterative optimization until the iteration termination condition is satisfied, determine the target power values corresponding to each energy unit, and schedule each energy unit based on the target power values, where the feasible region range includes the power upper limit value and the power lower limit value corresponding to each energy unit.

9. A terminal device, characterized in that, Comprising: One or more processors; A memory coupled to the processor for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the virtual power plant resource collaborative scheduling method according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the virtual power plant resource collaborative scheduling method according to any one of claims 1-7.

Images