CN-122022217-A - Virtual power plant and flexible load adjustment capacity prediction method for virtual power plant

CN122022217ACN 122022217 ACN122022217 ACN 122022217ACN-122022217-A

Abstract

The invention belongs to the technical field of flexible load adjustable capacity prediction, and particularly relates to a virtual power plant and a flexible load adjustable capacity prediction method of the virtual power plant. The method comprises the steps of determining influence factors influencing the electricity consumption of a flexible load user according to flexible load classification of the flexible load user, selecting historical data of the flexible load user on a non-response day, calculating a baseline load of the flexible load user based on the selected historical data on the non-response day, inputting the influence factor data of the flexible load user on a participation demand response day to be predicted into a flexible load prediction model to obtain a flexible load prediction result of the participation demand response day to be predicted, wherein the flexible load prediction model is obtained by training a machine learning model based on a historical flexible load value and corresponding historical influence factor data, and differentiating the flexible load prediction result with the baseline load to obtain a flexible load adjustment capacity prediction result of a virtual power plant, so that the flexible load adjustment capacity prediction accuracy of the virtual power plant is improved.

Inventors

Yang sihang
LUO KAIMING
LI YAHUI
CHEN YUXI
Dou Shangdie
Qu Xishuai
ZHANG HAO
LI ZHEN
XU JUN
YUAN JUNJUN
YANG TIANYE
WANG YIHANG
LI YONGLIANG
He Puxiang
MAO JIANRONG

Assignees

许昌许继软件技术有限公司
许继电气股份有限公司

Dates

Publication Date: 20260512
Application Date: 20241107

Claims (10)

1. A flexible load regulation capacity prediction method of a virtual power plant is characterized by comprising the following steps: According to the flexible load classification of the flexible load user, determining influence factors influencing the electricity load quantity of the flexible load user, and selecting historical data of the flexible load user on a non-response day; calculating a baseline load of the flexible load user based on the selected historical data of the non-response day; Inputting influence factor data of a flexible load user on a to-be-predicted participation demand response day into a flexible load prediction model to obtain a flexible load prediction result of the to-be-detected participation demand response day, wherein the flexible load prediction model is obtained by training a machine learning model based on a historical flexible load value and corresponding historical influence factor data; and (3) differentiating the flexible load prediction result with the baseline load to obtain a flexible load adjustment capability prediction result of the virtual power plant.
2. The method for predicting flexible load adjustment capacity of a virtual power plant according to claim 1, wherein when the flexible load type is a commercial building cold and hot load, the influencing factors include weather and temperature, and a working day, a rest day or a holiday, when the flexible load type is an electric car charging pile load, the influencing factors include weather, and a working day or a rest day, and when the flexible load type is a production equipment load, the influencing factors include equipment technical parameters and a production plan.
3. The method for predicting flexible load regulation capacity of a virtual power plant according to claim 2, wherein when the influence factors of the flexible load type include holidays, the users are classified by using a clustering analysis method, and then the baseline load of the flexible load users is calculated by using a mean method, a convolutional neural network algorithm or a weighted average method in the clustering algorithm.
4. The method for predicting flexible load adjustment capacity of a virtual power plant according to claim 3, wherein when the influence factors of the flexible load type include holidays and the baseline load of the flexible load user is calculated by using a weighted average method in a mean method or a clustering algorithm, the method for selecting the historical data of the flexible load user on a non-response day comprises selecting the historical load data of Q holidays before the participation demand response day in the non-response day if the participation demand response day is the holiday, selecting the historical load data of M weekdays before the participation demand response day in the non-response day if the participation demand response day is the weekday, and M > Q.
5. The method for predicting flexible load capacity of a virtual power plant according to claim 2, wherein when the flexible load type is a production equipment load, a weighted average method or a convolutional neural network algorithm is used to calculate a baseline load of a flexible load user.
6. The method for predicting the flexible load regulation capacity of the virtual power plant according to claim 3 or 5, wherein the method for calculating the baseline load of the flexible load user by adopting the convolutional neural network algorithm is characterized in that the convolutional neural network is trained based on historical electricity load data and each influence factor data, the influence factor data of the selected flexible load user on a non-response day is input into the trained convolutional neural network, and the baseline load of the flexible load user is output.
7. The method for predicting flexible load regulation capacity of a virtual power plant according to claim 1, wherein the selected historical influence factor data is data with higher similarity to actual influence factor data of a to-be-predicted participation demand response day, which is obtained by screening the historical influence factor data according to similarity when training a machine learning model.
8. The flexible load adjustment capacity prediction method of a virtual power plant according to claim 1 or 7, wherein the machine learning model is a Gaussian process regression model, and when the machine learning model adopts the Gaussian process regression model, the obtained flexible load prediction result of the day to be measured is a flexible load prediction interval, and the obtained flexible load adjustment capacity prediction result of the virtual power plant is an adjustment capacity prediction interval.
9. The method for predicting flexible load capacity of a virtual power plant as recited in claim 4, wherein Q is 5 and m is 10.
10. A virtual power plant comprising a processor, wherein the processor is configured to execute computer program instructions to implement the virtual power plant flexible load regulation capability prediction method of any of claims 1-9.

Description

Virtual power plant and flexible load adjustment capacity prediction method for virtual power plant Technical Field The invention belongs to the technical field of flexible load adjustable capacity prediction, and particularly relates to a virtual power plant and a flexible load adjustable capacity prediction method of the virtual power plant. Background When the supply and demand of the power system are in tension, the safe and stable operation of the power system is generally ensured by a demand response mode. The demand response is a means for enabling the power user to change the original power utilization mode by adjusting the price or giving an incentive to achieve the peak clipping purpose when the supply and the demand of the power system are in tension, so as to ensure the stable operation of the power system. The virtual power plant aggregates a large number of demand-side distributed flexible load resources that are spatially and spatially unevenly distributed and can participate as independent bodies in the power auxiliary service market or peak clipping demand response. At present, the demand response in China is mainly stimulated, the flexible load regulation capacity of the power consumer is generally predicted according to the electricity consumption experience value, the response load quantity of the demand response is determined according to the regulation capacity, then the response load quantity of the power consumer is declared to the virtual power plant to obtain economic compensation, and the virtual power plant participates in the electricity transaction market according to the flexible load response load quantity declared by the power consumer. However, the flexible load types are numerous, the flexible load adjustment capability predicted by the power user can have larger errors due to larger external influence, and the prediction error of the flexible load adjustment capability of the virtual power plant is larger. When the response load quantity declared by the power consumer is larger than the load quantity actually executed, the power grid dispatching is influenced, and meanwhile, the economic compensation obtained by the consumer is also smaller. Disclosure of Invention The invention aims to provide a virtual power plant and a virtual power plant flexible load adjustment capability prediction method, which are used for solving the problem of large virtual power plant flexible load adjustment capability prediction error in the prior art. The method comprises the steps of determining influence factors influencing the electricity load quantity of a flexible load user according to flexible load classification of the flexible load user, selecting historical data of the flexible load user on a non-response day, calculating a baseline load of the flexible load user based on the selected historical data of the non-response day, inputting the influence factor data of the flexible load user on a participation demand response day to be predicted into a flexible load prediction model to obtain a flexible load prediction result of the participation demand response day to be predicted, training the machine learning model based on a historical flexible load value and corresponding historical influence factor data by the flexible load prediction model, and differentiating the flexible load prediction result and the baseline load to obtain a flexible load adjustment capacity prediction result of a virtual power plant. Further, when the flexible load type is a commercial building cold and hot load, the influencing factors include weather and temperature, and working days, rest days or holidays, when the flexible load type is an electric car charging pile load, the influencing factors include weather, and working days or rest days, and when the flexible load type is a production equipment load, the influencing factors include equipment technical parameters and a production plan. Further, when the influence factors of the flexible load type include holidays, the users are classified by using a clustering analysis method, and then the baseline load of the flexible load users is calculated by using a mean method, a convolutional neural network algorithm or a weighted average method in the clustering algorithm. Further, when the influence factors of the flexible load type comprise holidays and the baseline load of the flexible load user is calculated by using a weighted average method in a mean method or a clustering algorithm, the method for selecting the historical data of the flexible load user on non-response days comprises the steps of selecting the historical load data of Q holidays before the participation demand response days in the non-response days if the participation demand response days are the holidays, selecting the historical load data of M working days before the participation demand response days in the non-response days if the participation demand response days are the working day