Power plant two-stage scheduling method fusing social influence graph and behavior response probability

Abstract

The invention relates to a power plant two-stage scheduling method fusing a social influence graph and a behavior response probability, the method comprises a day-ahead stage and an intra-day stage, the day-ahead stage collects historical operation data of distributed energy resources and user power consumption behavior data and carries out preprocessing, and a social influence graph model is constructed; a demand side response prediction result is generated in combination with the behavior response probability; based on the prediction result, constructing a multi-objective optimization scheduling model, generating an initial scheduling plan and decomposing the initial scheduling plan into a plurality of subtasks; in the intra-day stage, the change of the actual output of the distributed energy and the user power consumption behavior is monitored in real time, and the social influence graph model, the behavior response probability and the demand side response prediction result are updated; step-by-step correction is carried out on the sub-tasks in the day-ahead stage through a dynamic adjustment mechanism; and regulating and controlling the distributed energy resources and the energy storage equipment according to the corrected sub-tasks. Compared with the prior art, the method achieves the precise prediction of the power consumption behavior of the user at the demand side.

Description

Technical Field

[0001] This invention relates to the field of power system optimization scheduling technology, and in particular to a two-stage scheduling method for power plants that integrates social impact maps and behavioral response probabilities. Background Technology

[0002] With the deepening of energy transition, virtual power plants, as a new power system operation mode that integrates distributed energy resources, energy storage systems, and demand-side response, play a crucial role in improving energy utilization efficiency, promoting renewable energy consumption, and enhancing power system stability. Effective dispatching methods are essential for virtual power plants to achieve their functional objectives, among which the two-stage dispatching method has received widespread attention because it can optimize based on information and demand at different stages.

[0003] In existing technologies, there are various two-stage scheduling methods for virtual power plants. For example, some methods construct a two-layer optimization model based on the system flexibility boundary in the day-ahead phase, comprehensively considering the coordinated scheduling of grid-side flexibility resources and the day-ahead clearing of virtual power plants. In the real-time phase, they adjust the bidding schemes of virtual power plants and the active distribution network operation strategies to achieve the goals of minimizing the overall cost of the active distribution network and maximizing the market revenue of virtual power plants. Other methods formulate the optimal scheduling plan for the next day based on forecast information and sign an agreement with the day-ahead electricity market in the day-ahead phase. In the intraday phase, they use the day-ahead plan as a reference and adopt model predictive control strategies to adjust the intraday operation plan to eliminate net load fluctuations caused by forecast errors, while reducing penalties from the intraday balancing market and lowering overall operating costs. In addition, there are methods that establish an energy supply-side model, design the objective function and constraints for the optimal scheduling of virtual power plants under impact loads, form a day-ahead optimal scheduling plan based on the particle swarm optimization algorithm, and achieve power balance in intraday scheduling. The amount of controllable resources to be controlled within the day is determined based on relevant indicators, and the amount of controllable resources to be allocated to different equipment is further decomposed. At the same time, an intraday rolling optimization correction strategy based on model predictive control is adopted to deal with the power fluctuations of tie lines caused by prediction errors, and ensure that energy storage meets the daily operation energy balance constraints.

[0004] However, current two-stage scheduling methods for virtual power plants still have some technical limitations. Firstly, there are limitations in modeling distributed energy resources and user behavior. Distributed energy sources such as wind and solar power have significant uncertainties in output, significantly affected by natural factors like weather. Existing scheduling methods may have biases in capturing these changing patterns, thus affecting the accuracy of scheduling plans. Secondly, while demand response probabilities are considered for demand-side user behavior, they are usually only estimated in a simple way, without fully considering the complex social relationships between users and their impact on electricity consumption behavior. For example, within a community, users may imitate or influence each other, and existing models have not yet incorporated this social influence, resulting in insufficient accuracy in predicting user electricity consumption behavior, thus affecting the rationality and effectiveness of scheduling schemes. For instance, Chinese patent application CN120810609A discloses an optimized scheduling method for virtual power plants considering renewable energy, which considers the operating characteristics of virtual power plants and user electricity consumption behavior, constructs an intelligent demand response mechanism, and performs time-by-time supply and demand matching analysis between power generation control schemes and user response strategies, but it does not consider the social factors that may lead to mutual imitation or influence among users.

[0005] On the other hand, there is room for improvement in the coordination and optimization of the two-stage dispatch process. The connection between the day-ahead and intraday stages is not tight enough, and information transmission may be delayed and incomplete. The dispatch plan formulated in the day-ahead stage may not fully consider various unforeseen circumstances that may occur during the day and their dynamic impact on dispatch. When faced with significant deviations between actual and predicted wind and solar power output, or when user electricity demand fluctuates greatly due to sudden social events, intraday dispatch may struggle to make rapid, effective, comprehensive, and precise adjustments. This situation may affect the operational economy and stability of the virtual power plant. For example, when extreme weather causes a sharp drop in solar power output, the original two-stage dispatch plan may not be able to adjust the output of other energy resources in a timely and reasonable manner, potentially leading to insufficient power supply or increased system operating costs. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the existing technology by providing a two-stage power plant scheduling method that integrates social impact maps and behavioral response probabilities. This method enables accurate prediction of demand-side user electricity consumption behavior and effectively solves the problem of insufficient prediction accuracy in existing virtual power plant two-stage scheduling methods, which do not fully consider the social impact between users and only simply estimate behavioral response probabilities.

[0007] The objective of this invention can be achieved through the following technical solutions: A two-stage scheduling method for power plants that integrates social impact mapping and behavioral response probabilities, the method comprising: In the current phase: collect historical operational data of distributed energy resources and user electricity consumption behavior data and preprocess them; based on the user electricity consumption behavior data, construct a social impact graph model, quantify the interaction relationship between users, and generate demand-side response prediction results by combining behavioral response probabilities; based on the prediction results and historical operational data of distributed energy resources, construct a multi-objective optimization scheduling model that comprehensively considers economic and stability objectives, generate an initial scheduling plan and decompose it into multiple sub-tasks; Intraday Phase: Real-time monitoring of actual output of distributed energy resources and changes in user electricity consumption behavior; based on the changes in user electricity consumption behavior, updating the social impact diagram model and behavioral response probabilities, and recalculating the demand-side response forecast results; comparing and analyzing the recalculated demand-side response forecast results with the initial scheduling plan of the day-ahead phase, and using a rolling optimization strategy to progressively revise the sub-tasks of the day-ahead phase based on the actual output of distributed energy resources; regulating distributed energy resources and energy storage devices according to the revised sub-tasks.

[0008] Furthermore, the process of constructing the social impact graph model includes: Collect electricity consumption data of users within the community and extract correlation features between users, including similarity in electricity consumption behavior and / or social network relationships; Using users as nodes, the social impact graph model is constructed by connecting each node based on the associated features and determining the weight of the edges connecting each node by calculating the time series correlation of historical electricity consumption.

[0009] Furthermore, the similarity of electricity consumption behavior is calculated by analyzing the electricity consumption trends of different users within the same time period, and the social network relationship is obtained by investigating the actual social connections between users or by data analysis.

[0010] Furthermore, the behavioral response probability is generated by modeling the probability distribution of user electricity consumption behavior, and the modeling process includes: Collect users' historical electricity consumption data, analyze the distribution of each user's electricity consumption in different time periods, and obtain the minimum incentive threshold for electricity consumption of different types of users; Based on the preset incentive scheme, the probability distribution of each user's electricity consumption behavior in each time period is calculated, and the response probability of the current behavior is obtained according to the current time period.

[0011] Furthermore, the process of generating the demand-side response prediction results includes: The probability of behavioral response is obtained based on the probability distribution of the electricity consumption behavior; Based on the aforementioned behavioral response probability and social impact graph model, the electricity demand forecast results for users in the future time period are calculated through social communication relationships. The forecast results for all users are then aggregated to obtain the demand-side response forecast results.

[0012] Furthermore, when modeling the probability distribution of the user's electricity consumption behavior, the electricity consumption distribution of each user in different time periods follows a normal distribution.

[0013] Furthermore, the economic objectives of the multi-objective optimization scheduling model include minimizing the operating costs of the virtual power plant and maximizing market revenue. The stability objective of the multi-objective optimization scheduling model is constructed based on the matching degree between the output fluctuations of distributed energy resources and the electricity demand of different users. The constraints of the multi-objective optimization scheduling model include the upper and lower limits of the output of distributed energy resources, the charging and discharging limits of energy storage devices, and the degree to which user electricity demand is met.

[0014] Furthermore, the operating costs of the virtual power plant include the generation costs of distributed energy resources, the charging and discharging loss costs of energy storage devices, and the cost of purchasing electricity; the market revenue comes from the income obtained by the virtual power plant from selling excess electricity to the electricity market; the output fluctuation of the distributed energy resources is obtained by calculating the square of the deviation between the actual output and the predicted output of the distributed energy resources.

[0015] Furthermore, the constraint on the degree to which the user's electricity demand is met is obtained by constructing a user participation fatigue model.

[0016] Furthermore, during the regulation and control of the distributed energy resources and energy storage devices, the operating status data of the devices are collected in real time and fed back to the dispatch center. The dispatch center evaluates the execution effect of the dispatch instructions based on the feedback data and generates new dispatch instructions.

[0017] Compared with the prior art, the beneficial effects of the present invention include: 1. This invention constructs a social impact graph model that integrates the similarity of user electricity consumption behavior and social network relationships, and combines it with behavioral response probabilities generated by modeling based on the probability distribution of users' historical electricity consumption data. This enables accurate prediction of demand-side user electricity consumption behavior, effectively solving the problem of insufficient prediction accuracy in existing two-stage virtual power plant scheduling methods, which do not fully consider the social impact between users and only simply estimate behavioral response probabilities. This provides reliable data support for the rational formulation of subsequent scheduling plans and significantly improves the scientificity and adaptability of scheduling schemes.

[0018] 2. This invention dynamically updates the social impact diagram model and behavioral response probabilities by real-time monitoring of the actual output of distributed energy and changes in user electricity consumption behavior, and adopts a rolling optimization strategy to progressively correct sub-tasks. This breaks through the limitations of existing technologies, such as loose connection between the day-ahead and intraday stages, and delayed and incomplete information transmission. It ensures that the scheduling plan can be quickly and accurately adjusted when faced with sudden situations such as fluctuations in distributed energy output and sudden changes in user electricity demand, greatly enhancing the flexibility and anti-interference capability of virtual power plant scheduling.

[0019] 3. This invention designs a multi-objective optimization scheduling model to generate an initial scheduling plan, and ensures the effective execution of scheduling instructions through intraday dynamic adjustment and closed-loop control mechanisms. It effectively overcomes the problems of poor operational economy and low system stability caused by insufficient rationality of scheduling plans and untimely adjustments in existing methods. It achieves synergistic optimization of minimizing the operating cost of virtual power plants, maximizing market benefits, and improving the matching degree of distributed energy output and electricity demand, which significantly improves the overall operational efficiency and market competitiveness of virtual power plants. Attached Figure Description

[0020] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0022] Example 1 This invention discloses a two-stage scheduling method for power plants that integrates social impact maps and behavioral response probabilities. The method is as follows: Figure 1 As shown, steps S1-S4 are included, and the specific descriptions of each step are as follows: Step S1: Collect historical operation data of distributed energy resources and user electricity consumption behavior data and preprocess them; construct a social impact graph model based on user electricity consumption behavior data, quantify the interaction relationship between users, and generate demand-side response prediction results by combining behavioral response probabilities.

[0023] The process of constructing a social impact diagram model includes: Collect electricity consumption data of users in the community, extract the correlation features between users, and construct a peer set for each user. The correlation features include, but are not limited to, similarity of electricity consumption behavior and / or social network relationships. Using users as nodes, the system connects each node based on correlation features, and determines the weight of the edges connecting each node by calculating the time series correlation of historical electricity consumption, thus constructing a social impact graph model. Electricity consumption behavior similarity is calculated by analyzing the electricity consumption trends of different users within the same time period, that is, comparing the similarity of the electricity consumption curves of two users within the same time period. Social network relationships are derived by investigating the actual social connections between users or by data analysis.

[0024] The weights of the edges connecting each node are determined by calculating the time series correlation of historical electricity consumption. Specifically, the influence strength of the current user's peers within the peer set is fitted using the historical demand response data of each user.

[0025] Therefore, the construction process of the social influence graph model is based on users' social relationships, clarifying the peer group corresponding to each user; combining users' historical demand response data, analyzing the strength of each user's influence by their peers, and generating a social influence graph with users as nodes and the influence strength between peers as edges, which can quantify the interaction relationship between users.

[0026] The behavioral response probability is generated by modeling the probability distribution of user electricity consumption behavior. The specific process is as follows: By combining users' historical incentive data, we can determine the minimum incentive threshold for different types of users to participate in the response, i.e., electricity consumption. Then, based on the current preset incentive scheme, we can analyze the probability of each user participating in the demand response at different times.

[0027] When modeling the probability distribution of user electricity consumption behavior, historical electricity consumption data of users is collected, and the electricity consumption distribution of each user in different time periods is statistically analyzed. The electricity consumption follows a normal distribution.

[0028] The process of generating demand-side response forecast results includes: The probability of the behavior response is obtained based on the probability distribution model generated when obtaining the probability of the behavior response above. Based on the behavioral response probability and social impact graph model, the electricity demand forecast of users in the future time period is calculated through social communication relationships. Specifically, the social impact graph model can obtain each user's autonomous response intention, the response behavior of their peer group and the corresponding influence intensity. Based on these values, the electricity demand forecast of users in the future time period is calculated, and the forecast results of all users are summarized to obtain the demand-side response forecast result.

[0029] In the process of generating demand-side response forecasts, the subjective value of users to the current incentives is also evaluated (distinguishing between perceived gains and losses, ensuring that the perceived gains are not less than the losses), and the incentive scheme is adjusted to a reasonable range.

[0030] Specifically, in this embodiment, in the first stage, i.e., the current stage, electricity consumption data of users within the community is collected to extract correlation features between users. These features include, but are not limited to, electricity consumption behavior similarity and social network relationships. Electricity consumption behavior similarity can be calculated by analyzing the electricity consumption trends of different users within the same time period. For example, comparing the electricity consumption curves of user A and user B over a consecutive week, the correlation coefficient between the two curves is calculated as a similarity index. Social network relationships can be obtained by investigating the actual social connections between users or through data analysis, such as whether users belong to the same family, the same community, or have shared activity records. This data is input into a preset social influence graph model, which uses user nodes as the core and uses edge weights to represent the degree of mutual influence between users. The edge weights are calculated from the correlation of electricity consumption behavior in historical data. For example, by calculating the time series correlation of electricity consumption of user A and user B over the past month, the weight value between them is 0.85. This weight value reflects the strength of the potential influence of user A on user B's electricity consumption behavior.

[0031] Step S2: Based on the day-ahead forecast results and historical operating data of distributed energy resources, construct a multi-objective optimization scheduling model that comprehensively considers economic and stability objectives, generate an initial scheduling plan and decompose it into multiple sub-tasks.

[0032] In this embodiment, the objective function includes two aspects: economic efficiency and stability. The economic objective primarily focuses on minimizing the operating costs of the virtual power plant and maximizing market revenue. Operating costs include the generation costs of distributed energy resources, the charging and discharging losses of energy storage devices, and the cost of purchasing electricity. Market revenue comes from the income obtained by the virtual power plant from selling excess electricity to the electricity market. The stability objective focuses on the degree of matching between the output fluctuations of distributed energy resources and user electricity demand. For example, it assesses the system's stability level by calculating the sum of squared deviations between the actual and predicted output of distributed energy resources. Constraints include upper and lower limits on the output of distributed energy resources, charging and discharging limits on energy storage devices, and the degree to which user electricity demand is met.

[0033] The objective function can be expressed in the following form: in, The weighting of power generation costs, i.e., the operating costs of virtual power plants, Weights for behavioral utility The weights for the scheduling mismatch penalty are the stability objective. For scheduling periods, For scenarios involving resource scheduling, For power distribution nodes, For the scene The amount of resources to be dispatched, such as the dispatch amount or planned power generation and installed capacity of photovoltaic, wind power, and energy storage. For nodes The marginal cost of flexible power generation dispatch, for Time Period Scene Next The basic power generation of each node Based on user types, users are categorized into different electricity consumption types according to complex psychological and social effects, and there are influences between users. For users participating in demand response, For the first Risk attitude of individual user types For users During the period Internal demand response volume For the first The prospective value function for each user type quantifies the user's subjective perception of gain / loss. For user gains / losses, Let be the penalty coefficient for positive scheduling deviation under the nth node. For node n and time period Scene The positive deviation in scheduling, i.e., the positive deviation between the actual output and the predicted output of distributed energy resources, is the quantified positive deviation between these two values. This is the penalty coefficient for negative scheduling deviations at the nth node. For node n and time period Scene The negative deviation in scheduling refers to the negative deviation between the actual output and the predicted output of distributed energy resources. in, For type The curvature parameter of the value function when users receive benefits. For type The curvature parameter of the value function when the user loses For type User loss aversion coefficient The larger the user, the more sensitive they are to the loss. in, For user u during the time period The individual's willingness to respond, i.e., the tendency to respond autonomously when not influenced by peers. For type A user's social sensitivity index measures the degree to which a user is influenced by their peers. The strength weight of the influence of user u on peer v. For companion v in time period Demand response volume, Let be the set of users u's peers, that is, other users who can influence user u.

[0034] By adjusting , and Given this value, decision-makers can design strategies that align with policy priorities, such as cost minimization, weighting based on user type, or adjusting scheduling reliability under conditions of uncertainty.

[0035] By improving and The numerical value can identify nodes with smaller stability margins or higher load sensitivity, allowing system administrators to formulate more conservative scheduling schemes when necessary.

[0036] The constraint condition can be expressed in the following form: Upper and lower limits of output from distributed energy resources: in, For node n in time period Scene The energy storage discharge capacity at that time For node n in time period Scene The energy storage charging capacity below, Let n be the set of users covered by node n. This refers to the set of allocation nodes, i.e., the nodes participating in energy allocation in the scheduling system, such as community / building-level scheduling nodes. For user u during the time period Scene The following demand response adjustment amount, For node n in time period Scene The base load, which represents the user's basic electricity demand, is a net consumption item. For node n in time period Scene The energy imbalance is used to match supply and demand deviations and is a regulating term to maintain energy consistency.

[0037] The above constraint expression is in the scenario Below, all allocated nodes During the period The core node energy balance constraint. It ensures that net generation (including discharge storage and behavioral response) minus net consumption (including charging and baseload) must equal the total imbalance. Matching. This is the operational glue that maintains energy consistency in every VPP scheduling scenario.

[0038] Where m is the node adjacent to n, These are the elements of the admittance matrix between node n and its adjacent node m, representing the electrical connection characteristics between nodes n and m, and are the core coefficients for voltage constraint linearization. For node n in time period Scene The node voltage, i.e., the actual voltage value of a node in a power distribution system. For node n, the neighboring node m in the time period Scene The node voltage below, This represents the maximum permissible voltage deviation at node n.

[0039] The above constraint expression uses the admittance matrix. To implement linearized node voltage constraints, ensure that the voltage difference between adjacent nodes remains within allowable deviation ranges. This is achieved by setting a maximum power limit. Even in situations where there is uncertainty in the behavior of the dispatch, it can ensure that the power quality remains within the regulatory range.

[0040] Furthermore, the constraints on the upper and lower limits of distributed energy resource output also include physical limitations on the operational boundaries of power generation and energy storage systems at nodes, ensuring that they can operate in any scenario. Below, at any time The amount of electricity that can be generated, charged into storage, or discharged is physically limited.

[0041] The charging and discharging limits of energy storage devices are calculated based on actual user electrical equipment using existing physical formulas.

[0042] The degree to which users' electricity needs are met can be expressed using a participation fatigue model, which states that after a period of use, users will enter a rest phase, as shown in the following expression: in, For users During the period Internal demand response volume For type The user's engagement fatigue decay coefficient is used to capture the user's need for rest.

[0043] The solution to the above model is based on existing technology, which will not be elaborated here.

[0044] A specific application example: A photovoltaic power generation system has a maximum output of 50 kW and a minimum output of 0 kW; the energy storage device has a maximum charging power of 20 kW and a maximum discharging power of 30 kW; the user's electricity demand must be fully met within each time period. The above model is solved using existing methods to generate an initial scheduling plan, which is then decomposed into multiple sub-tasks. Each sub-task corresponds to a specific time period and a set of distributed energy resources. For example, the task for the period from 8:00 AM to 10:00 AM is responsible for scheduling the operation status of the photovoltaic power generation system and the energy storage device.

[0045] Step S3: During the daytime phase, a dynamic adjustment mechanism is introduced to monitor the actual output of distributed energy and changes in user electricity consumption behavior in real time. Based on the changes in user electricity consumption behavior, the social impact diagram model and behavioral response probability are updated, and the demand-side response prediction results are recalculated. The recalculated demand-side response prediction results are compared and analyzed with the initial scheduling plan of the daytime phase. Based on the actual output of distributed energy, a rolling optimization strategy is adopted to progressively revise the sub-tasks of the daytime phase.

[0046] The specific process of revising the sub-tasks of the rolling optimization of the intraday scheduling plan is as follows: the objective function of the scheduling is revised based on real-time data, with a focus on adjusting the penalty cost of supply and demand imbalance; the energy balance and node voltage constraints are re-verified, and the deviation between actual operation and plan is made up by adjusting the charging and discharging of energy storage devices and the output of distributed power sources; finally, real-time revised scheduling plans for each time period are generated.

[0047] Specifically, the dynamic adjustment mechanism is used to address unforeseen circumstances that may arise during actual operation. First, by monitoring changes in user electricity consumption behavior in real time, the edge weights and behavioral response probabilities in the social impact graph model are updated. For example, if user D's actual electricity consumption on a certain day is significantly lower than the predicted value, the edge weights between user D and other users need to be recalculated. Second, based on the updated model, the demand-side response forecast is recalculated and compared with the day-ahead scheduling plan. For example, if user E's actual electricity consumption is 10% higher than the predicted value, the overall electricity demand forecast for their community needs to be adjusted. Next, a rolling optimization strategy is used to progressively correct sub-tasks. This strategy uses the current actual data as a basis, combined with forecast data for the next period, and re-optimizes the scheduling scheme for sub-tasks based on a multi-objective optimization scheduling model. For example, at 10:00 AM, the system re-optimizes the task scheduling scheme for the period from 10:00 AM to 12:00 PM based on the current actual output data and the forecast data for the next two hours. In this way, the deviation between the day-ahead scheduling plan and the actual operating status is gradually reduced.

[0048] Step S4: Regulate distributed energy resources and energy storage devices according to the revised sub-tasks.

[0049] During the regulation and control of distributed energy resources and energy storage devices, real-time operational status data of the devices is collected and fed back to the dispatch center. The dispatch center evaluates the execution effect of dispatch instructions based on the feedback data and generates new dispatch instructions. Specifically, the regulation and control involves converting the intraday revised dispatch plan into equipment regulation instructions, including adjusting the output of distributed power sources, the charging and discharging status of energy storage devices, and the demand response of users, while simultaneously updating the remaining power status of electric vehicle batteries in real time.

[0050] Specifically, dispatch instructions are transmitted to the execution module via a communication network. The execution module then regulates distributed energy resources and energy storage devices according to the instructions. For example, a dispatch instruction might require the photovoltaic power generation system to maintain maximum output from 10:00 AM to 12:00 PM, while the energy storage device charges at a power of 20 kilowatts. Upon receiving the instruction, the execution module sends control signals to both the photovoltaic power generation system and the energy storage device to adjust their operating status. During the regulation process, the execution module collects real-time operating status data of the devices, such as the actual output data of the photovoltaic power generation system and the current power status of the energy storage device, and feeds this data back to the dispatch center.

[0051] The dispatch center evaluates the effectiveness of dispatch instructions based on feedback data and generates new instructions when necessary. For example, if the actual charging power of energy storage devices is lower than expected, a new instruction is generated to adjust the charging rate. More specifically, during execution, the actual operating status of the devices and the actual response data of users are continuously collected to evaluate the effectiveness of dispatch instructions. If problems such as excessive voltage deviation or excessive supply-demand imbalance are found, the system immediately returns to the intraday behavior response update stage to readjust the dispatch plan. This closed-loop control mechanism effectively improves the flexibility and adaptability of the dispatch process.

[0052] In the method of this invention, the output of the social impact graph model is directly used as input data for the demand-side response prediction algorithm. The demand-side response prediction results are then used to guide the design of a multi-objective optimization scheduling model. The output of the multi-objective optimization scheduling model, i.e., the initial scheduling plan, is decomposed into multiple sub-tasks and then entered into a dynamic adjustment mechanism for real-time correction. The output of the dynamic adjustment mechanism is transmitted to the execution module through a multi-level instruction distribution and feedback mechanism, ultimately achieving precise control of distributed energy resources and energy storage devices. For example, when the actual output of the photovoltaic power generation system is lower than the predicted value, the dynamic adjustment mechanism will recalculate the demand-side response prediction results and generate new scheduling instructions, which are then transmitted to the execution module through the instruction distribution and feedback mechanism to adjust the discharge power of the energy storage devices to compensate for the deficiencies of the photovoltaic power generation system.

[0053] Example 2 Based on Embodiment 1, this embodiment provides an electronic device, including: one or more processors and a memory, wherein the memory stores one or more programs, the one or more programs including instructions for executing the power plant two-stage scheduling method as described above that integrates social impact maps and behavioral response probabilities.

[0054] At the hardware level, the electronic device includes a processor, internal bus, network interface, memory, and non-volatile memory, and may also include other hardware required for business operations. The processor reads the corresponding computer program from the non-volatile memory into memory and then runs it to implement the aforementioned two-stage scheduling method for power plants that integrates social impact maps and behavioral response probabilities. Of course, in addition to software implementation, this invention does not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. That is to say, the execution entity of the following processing flow is not limited to individual logic units, but can also be hardware or logic devices.

[0055] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0056] Computer-readable media include both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0057] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A two-stage scheduling method for power plants that integrates social impact maps and behavioral response probabilities, characterized in that, The method includes: In the current phase: collect historical operational data of distributed energy resources and user electricity consumption behavior data and preprocess them; based on the user electricity consumption behavior data, construct a social impact graph model, quantify the interaction relationship between users, and generate demand-side response prediction results by combining behavioral response probabilities; based on the prediction results and historical operational data of distributed energy resources, construct a multi-objective optimization scheduling model that comprehensively considers economic and stability objectives, generate an initial scheduling plan and decompose it into multiple sub-tasks; Intraday Phase: Real-time monitoring of actual output of distributed energy resources and changes in user electricity consumption behavior; based on the changes in user electricity consumption behavior, updating the social impact diagram model and behavioral response probabilities, and recalculating the demand-side response forecast results; comparing and analyzing the recalculated demand-side response forecast results with the initial scheduling plan of the day-ahead phase, and using a rolling optimization strategy to progressively revise the sub-tasks of the day-ahead phase based on the actual output of distributed energy resources; regulating distributed energy resources and energy storage devices according to the revised sub-tasks.

2. The two-stage scheduling method for power plants integrating social impact maps and behavioral response probabilities as described in claim 1, characterized in that, The process of constructing the social impact graph model includes: Collect electricity consumption data of users within the community and extract correlation features between users, including similarity in electricity consumption behavior and / or social network relationships; Using users as nodes, the social impact graph model is constructed by connecting each node based on the associated features and determining the weight of the edges connecting each node by calculating the time series correlation of historical electricity consumption.

3. The two-stage scheduling method for power plants integrating social impact maps and behavioral response probabilities as described in claim 2, characterized in that, The similarity of electricity consumption behavior is calculated by analyzing the electricity consumption trends of different users within the same time period, and the social network relationship is obtained by investigating the actual social connections between users or by data analysis.

4. The two-stage scheduling method for power plants integrating social impact maps and behavioral response probabilities as described in claim 1, characterized in that, The behavioral response probability is generated by modeling the probability distribution of user electricity consumption behavior, and the modeling process includes: Collect users' historical electricity consumption data, analyze the distribution of each user's electricity consumption in different time periods, and obtain the minimum incentive threshold for electricity consumption of different types of users; Based on the preset incentive scheme, the probability distribution of each user's electricity consumption behavior in each time period is calculated, and the response probability of the current behavior is obtained according to the current time period.

5. The two-stage scheduling method for power plants integrating social impact maps and behavioral response probabilities as described in claim 4, characterized in that, The process of generating the demand-side response prediction results includes: The probability of behavioral response is obtained based on the probability distribution of the electricity consumption behavior; Based on the aforementioned behavioral response probability and social impact graph model, the electricity demand forecast results for users in the future time period are calculated through social communication relationships. The forecast results for all users are then aggregated to obtain the demand-side response forecast results.

6. The two-stage scheduling method for power plants integrating social impact maps and behavioral response probabilities according to claim 4, characterized in that, When modeling the probability distribution of the user's electricity consumption behavior, the electricity consumption distribution of each user in different time periods follows a normal distribution.

7. The two-stage scheduling method for power plants integrating social impact maps and behavioral response probabilities as described in claim 1, characterized in that, The economic objectives of the multi-objective optimization scheduling model include minimizing the operating costs of the virtual power plant and maximizing market revenue. The stability objective of the multi-objective optimization scheduling model is constructed based on the matching degree between the output fluctuations of distributed energy resources and the electricity demand of different users. The constraints of the multi-objective optimization scheduling model include the upper and lower limits of the output of distributed energy resources, the charging and discharging limits of energy storage devices, and the degree to which user electricity demand is met.

8. The two-stage scheduling method for power plants integrating social impact maps and behavioral response probabilities according to claim 7, characterized in that, The operating costs of the virtual power plant include the generation costs of distributed energy resources, the charging and discharging losses of energy storage devices, and the cost of purchasing electricity; the market revenue comes from the income obtained by the virtual power plant from selling excess electricity to the electricity market; the output fluctuation of the distributed energy resources is obtained by calculating the square of the deviation between the actual output and the predicted output of the distributed energy resources.

9. A two-stage power plant scheduling method integrating social impact maps and behavioral response probabilities as described in claim 7, characterized in that, The constraint on the degree to which user electricity demand is met is obtained by constructing a user participation fatigue model.

10. A two-stage power plant scheduling method integrating social impact maps and behavioral response probabilities as described in claim 1, characterized in that, During the regulation and control of the distributed energy resources and energy storage devices, the operating status data of the devices are collected in real time and fed back to the dispatch center. The dispatch center evaluates the execution effect of the dispatch instructions based on the feedback data and generates new dispatch instructions.

Images