CN-122026439-A - Power plant two-stage scheduling method integrating social influence graph and behavior response probability

CN122026439ACN 122026439 ACN122026439 ACN 122026439ACN-122026439-A

Abstract

The invention relates to a power plant two-stage scheduling method integrating a social influence graph and behavior response probability, which comprises a day-ahead stage and a day-in stage, wherein the day-ahead stage collects historical operation data of distributed energy resources and user electricity behavior data and performs preprocessing to construct a social influence graph model, and a demand side response prediction result is generated by combining the behavior response probability; the method comprises the steps of constructing a multi-target optimized scheduling model based on a prediction result, generating an initial scheduling plan and decomposing the initial scheduling plan into a plurality of subtasks, monitoring the actual output of distributed energy and the change of electricity consumption behaviors of users in real time in a daily stage, updating a social influence graph model, behavior response probability and a demand side response prediction result, correcting the subtasks in a daily stage step by step through a dynamic adjustment mechanism, and regulating and controlling distributed energy resources and energy storage equipment according to the corrected subtasks. Compared with the prior art, the method and the device realize accurate prediction of the electricity consumption behavior of the user at the demand side.

Inventors

LU YI
DONG YUFANG
SHENG MING
LIU ZITENG
ZHAO JIANLI
WANG QING
ZHENG QINGRONG
HUANG FENG
Zhu Shana
ZHANG RUI

Assignees

国网上海市电力公司

Dates

Publication Date: 20260512
Application Date: 20251216

Claims (10)

1. A power plant two-stage scheduling method integrating social influence graph and behavior response probability is characterized by comprising the following steps: The method comprises the steps of acquiring historical operation data of distributed energy resources and user electricity behavior data and preprocessing the historical operation data, constructing a social influence graph model based on the user electricity behavior data, quantifying interaction relation among users, and generating a demand side response prediction result by combining behavior response probability; The method comprises the steps of monitoring actual output of distributed energy and change of user electricity consumption behavior in real time, updating a social influence graph model and behavior response probability based on the change of the user electricity consumption behavior, recalculating a demand side response prediction result, comparing and analyzing the recalculated demand side response prediction result with an initial scheduling plan in the day-ahead stage, correcting subtasks in the day-ahead stage step by step based on the actual output of the distributed energy by adopting a rolling optimization strategy, and regulating and controlling distributed energy resources and energy storage equipment according to the corrected subtasks.
2. The two-stage scheduling method of a power plant fusing social influence graph and behavior response probability as set forth in claim 1, wherein the process of constructing the social influence graph model includes: collecting electricity consumption data of users in communities, and extracting association features among the users, wherein the association features comprise electricity consumption behavior similarity and/or social network relations; And connecting the nodes based on the association characteristics by taking the users as nodes, determining the weight of the edge connected with the nodes by calculating the time sequence correlation of the historical electricity consumption, and constructing the social influence graph model.
3. The power plant two-stage scheduling method integrating social influence graphs and behavior response probabilities according to claim 2, wherein the electricity consumption behavior similarity is calculated by analyzing electricity consumption change trends of different users in the same time period, and the social network relationship is obtained by investigating actual social connection or data analysis among the users.
4. The two-stage scheduling method of a power plant fusing social influence graph and behavior response probabilities of claim 1, wherein the behavior response probabilities are generated by modeling probability distributions of user power consumption behaviors, the modeling process comprising: collecting historical electricity consumption data of users, counting electricity consumption distribution conditions of each user in different time periods, and obtaining the lowest excitation threshold of electricity consumption of different types of users; based on a preset excitation scheme, calculating probability distribution of electricity consumption behaviors of each user in each period, and acquiring current behavior response probability according to the current period.
5. The two-stage scheduling method of a power plant integrating social influence graph and behavioral response probabilities of claim 4, wherein generating the demand-side response prediction result comprises: acquiring behavior response probability based on the probability distribution of the electricity utilization behavior; And calculating the electricity demand prediction results of the users in the future time period through the social propagation relationship based on the behavior response probability and the social influence graph model, and summarizing the prediction results of all the users to obtain a demand side response prediction result.
6. The two-stage scheduling method for a power plant, which combines a social influence graph and behavior response probabilities, according to claim 4, wherein when modeling the probability distribution of the user electricity behavior, the electricity consumption distribution of each user in different time periods obeys normal distribution.
7. The two-stage scheduling method of a power plant, which combines a social influence graph and a behavior response probability, according to claim 1, wherein the economic targets of the multi-target optimal scheduling model comprise the minimization of the running cost and the maximization of market income of a virtual power plant, the stability targets of the multi-target optimal scheduling model are constructed based on the matching degree of the output fluctuation of distributed energy resources and the power consumption demands of different users, and the constraint conditions of the multi-target optimal scheduling model comprise the upper limit and the lower limit of the output of the distributed energy resources, the charge and discharge limitation of energy storage equipment and the satisfaction degree of the power consumption demands of the users.
8. The two-stage scheduling method for the power plant, which combines the social influence graph and the behavior response probability, according to claim 7, is characterized in that the running cost of the virtual power plant comprises the power generation cost of the distributed energy resources, the charge and discharge loss cost of the energy storage equipment and the electricity purchasing cost, the market benefit is derived from the income obtained by selling surplus electric quantity to the electric market by the virtual power plant, and the output fluctuation of the distributed energy resources is obtained by calculating the square of deviation between the actual output and the predicted output of the distributed energy resources.
9. The two-stage scheduling method for the power plant, which integrates the social influence graph and the behavior response probability, according to claim 7, wherein the constraint of the satisfaction degree of the user electricity demand is obtained by constructing a participation fatigue model of the user.
10. The two-stage scheduling method of the power plant integrating the social influence graph and the behavior response probability according to claim 1, wherein in the process of regulating and controlling the distributed energy resources and the energy storage equipment, the running state data of the equipment are collected in real time and fed back to a scheduling center, and the scheduling center evaluates the execution effect of the scheduling instruction according to the feedback data to generate a new scheduling instruction.

Description

Power plant two-stage scheduling method integrating social influence graph and behavior response probability Technical Field The invention relates to the technical field of power system optimization scheduling, in particular to a power plant two-stage scheduling method integrating social influence graphs and behavior response probabilities. Background Along with the deep propulsion of energy transformation, the virtual power plant is used as a novel power system operation mode integrating distributed energy resources, an energy storage system and a demand side response, and plays a key role in improving energy utilization efficiency, promoting renewable energy consumption, improving power system stability and the like. The effective scheduling method is important for realizing the functional target of the virtual power plant, wherein the two-stage scheduling method is widely focused because the two-stage scheduling method can be optimized by combining information and requirements of different stages. In the prior art, there are a number of two-stage scheduling methods for virtual power plants. For example, part of the method is based on a system flexibility boundary in a day-ahead stage, a double-layer optimization model which comprehensively considers network side flexibility resource coordination scheduling and day-ahead clearance of the virtual power plant is constructed, and a virtual power plant bidding scheme and an active power distribution network operation strategy are adjusted in a real-time stage, so that the aims of minimizing the comprehensive cost of the active power distribution network and maximizing the market benefit of the virtual power plant are achieved. The method also comprises the steps of making an optimal scheduling plan of the next day according to the prediction information in the day-ahead stage, signing an agreement with the day-ahead electric power market, taking the day-ahead plan as a reference in the day-ahead stage, and adopting a model prediction control strategy to adjust the operation plan in the day so as to eliminate net load fluctuation caused by prediction errors, reduce fines from the day-ahead balance market and reduce the overall operation cost. In addition, there is also a method for forming a daily optimization scheduling plan based on a particle swarm algorithm by establishing an energy supply side model, designing an objective function and constraint conditions of virtual power plant optimization scheduling under impact load, determining a regulating and controlling quantity of daily controllable resources based on related indexes for the purpose of power balance, further decomposing the regulating and controlling quantity to be distributed to different devices, and simultaneously adopting a daily rolling optimization correction strategy of model prediction control to cope with tie line power fluctuation caused by prediction errors, so that energy storage is ensured to meet daily operation energy balance constraint. However, there are still some technical pain points in these two-stage scheduling methods of virtual power plants. On the one hand, there is a limitation in modeling distributed energy resources and user behavior. The output of distributed energy sources such as wind power and photovoltaic has stronger uncertainty, is obviously influenced by natural factors such as weather and the like, and the existing scheduling method can have deviation when capturing the change rule, thereby influencing the accuracy of a scheduling plan. Meanwhile, for the behavior response of the user at the demand side, although the demand response probability is considered, only simple estimation is generally performed, and the complex social relationship among the users and the influence of the complex social relationship on the electricity consumption behavior are not fully considered. For example, in a community, users may have mutual imitations or influences, but the existing model has not included such social influence factors, so that the prediction accuracy of the electricity consumption behavior of the users is insufficient, and the rationality and the effectiveness of the scheduling scheme are further influenced. For example, chinese patent application CN120810609a discloses a virtual power plant optimizing and scheduling method considering renewable energy, which considers the operation characteristics of the virtual power plant and the electricity consumption behavior of users, constructs an intelligent demand response mechanism, and performs a time-period supply-demand matching analysis on the power generation regulation scheme and the user response strategy, but does not consider social factors that may have mutual imitations or influences between users. On the other hand, a certain lifting space exists in the aspects of coordination and optimization in the two-stage scheduling process. The connection between the pre-day stage and the intra-day stage is not tig