基于确定性退火聚类的LSSVM短期负荷预测

Short-term Load Forecasting Based on LSSVM and Deterministic Annealing Clustering

提出了确定性退火聚类和最小二乘支持向量机(Least Square Support Vectorma-chine,LSSVM)相结合的电力系统短期负荷预测方法。考虑影响负荷变化的各种因素构造负荷样本数据,利用确定性退火聚类算法对样本数据进行分类,得到的分类样本数据作为最小二乘支持向量机的学习样本,保证最小二乘支持向量机具有较高的预测精度。利用某电力公司2007年负荷数据和气象数据进行仿真实验,仿真结果表明该方法具有较高的预测精度。

A short-term load forecasting method based on LSSVM and deterministic annealing clustering algorithm is proposed. All kinds of factors affecting load data are considered to form load samples. Load samples are classified using deterministic annealing algo- rithm, then fed into the least square support vector machines. The load data and meteorological data of a electrical company in 2007 is utilized to test the forecasting model. The simulation result shows that the proposed method can improve the predicting accuracy.