分时电价下用户响应行为的模型与算法  

Model and Algorithm of Customers' Responsive Behavior Under Time-of-Use Price

作　　者：刘继东[1] 韩学山[1] 韩伟吉[1] 张利[1] 

机构地区：[1]电网智能化调度与控制教育部重点实验室(山东大学),山东省济南市250061

出　　处：《电网技术》2013年第10期2973-2978,共6页Power System Technology

基　　金：国家自然科学基金项目(51077087);山东省自然科学基金项目(Y2008F19)~~

摘　　要：为满足需求响应机制中描述用户行为规律的需要,提出一种电力用户需求响应行为的模型与算法。在获取足够的用户历史数据的基础上,通过支持向量机(support vector machine,SVM)回归进行数据挖掘,建立了电力用户在分时电价下的响应行为模型。该方法以用户响应的影响因素分析为基础,确定了回归模型的输入与输出属性;并通过定义等效电价比率,构建了含丰富数据信息的训练样本;最后采用网格搜索法选择SVM回归的最佳参数,实现了回归模型的高精度预测。该模型实现了电力用户在分时电价下行为规律的模拟,可揭示用户响应电量变化与分时电价政策激励力度间的关系,从而为更多研究提供基础数据。仿真分析证明了该模型和算法的有效性和合理性。To meet the demand of describing power customer＇s behavior rule in demand response mechanism, a model and an algorithm for demand response behavior of power customer are proposed. On the basis of acquiring enough customer historical data, the data mining is performed by support vector machine（SVM） regression to build customer＇s response behavior model under time-of-use（TOU） price. In the presented method, firstly based on the analysis on impacting factors of customer＇s response the input and output attributes of the regression model are determined; then through defining equivalent price ratio the training sample containing plenty of data information is constructed; finally optimal parameters for SVM regression are chosen by grid search method（GSM）, thus high-precision prediction of regression model is implemented. Using the proposed model the behavior rule of power customer under TOU price is simulated, by which the relation between the electricity quantity variation of customer response and incentive strength of TOU price policy can be revealed and pursuantly basic data can be offered for researches in future.

关 键 词：需求响应 分时电价 用户行为 支持向量机 回归分析 

分 类 号：TM714[电气工程—电力系统及自动化]