Electric vehicle standby service load aggregation method based on multi-arm learning machine

Abstract

The invention discloses an electric vehicle standby service load aggregation method based on a multi-arm learning machine, and belongs to the field of power system demand responding. The method comprises the steps that: firstly, a behavior rocker model of an electric vehicle user responding to a load aggregation request signal and a load aggregation target function are established according to the mode that an electric vehicle participates in a standby service, and then an electric vehicle standby service load aggregation model is established; and then, electric vehicle user behavior rocker arm model characteristics are considered, a load aggregation target is weighed based on risk avoidance, and an electric vehicle user selection algorithm based on risk avoidance is proposed to learn various response behaviors of each user and continuously update estimation of the user behaviors, so that a proper user is selected to participate in the standby service in a proper mode, and a load aggregation target is completed. According to the method, a more reliable load aggregation effect, lower load aggregation cost and higher user satisfaction can be obtained, and an effective way is provided for an aggregator to balance the load aggregation reliability and the load aggregation cost.

Description

technical field [0001] The invention belongs to the field of power system demand response load aggregation control, and more specifically relates to a load aggregation method for electric vehicle backup service based on a multi-arm learning machine. Background technique [0002] At present, traditional low-efficiency coal-fired units are gradually withdrawing from the power system, and the proportion of new energy is gradually increasing. The output of new energy is highly random and uncertain, and large-scale new energy grid integration poses a huge challenge to the reliable operation and optimization of the power system. This requires the system to be equipped with more flexible reserve capacity to deal with the uncertainty risk caused by the forecast error of new energy output. However, the traditional thermal power units that can provide spinning backup are gradually decreasing, and new energy units do not have the ability to provide backup. Therefore, there is insuffi...

AI Technical Summary

Problems solved by technology

However, in the actual load aggregation process, there has always been the problem of uncertainty in the response behavior of electric vehicle users. Users may withdraw from demand response midway due to their own energy needs, travel arrangements, living habits, user fatigue, etc.

Unknown and uncertain user response behavior may cause extremely unreliable load aggregation effects. For example, if a large number of users quit demand response midway, the load aggregation amount will seriously deviate from the target amount, resulting in unreliable backup services provided by load aggregation.

At present, the research on the load aggregation method of electric vehicles focuses on the consideration of the aggregator's economic benefits, user satisfaction and other factors to formulate a demand response mechanism, but few studies start from exploring the characteristics of user behavior, and fail to essentially solve the problem of user behavior uncertainty. , there is a lack of reliable load aggregation methods for EV standby services

Images