电力系统供电短期负荷预测方法仿真研究

摘要

It can guarantee safe and stable operation of power grid to predict power supply short-term load of power system accurately.Traditional method has disadvantage of great prediction error and poor real-time.In this paper,we propose a new prediction method.It divides area users into six clusters via the K-means data mining (DM) model.It makes the same cluster sample have similarity and improves the prediction accuracy.In the input variable of prediction model,we takes natural environment and historical load into account.Finally,we integrate the modified Cuckoo Search Support Vector Machine (ICSSVM) to get itemized prediction,and the last prediction result is obtained.The simulation results show that above model has more accurate prediction results than the traditional method.%针对电力系统供电短期负荷进行准确预测为了保证电网安全稳定的运行.但传统方法进行电力系统供电短期负荷预测存在预测误差大、实时性差的弊端.提出一种新的电力系统供电短期负荷预测方法,利用K-means数据挖掘模型将区域用户划分为6个群组,使得同一群体中的样本具有相似性,提高预测精度.在预测模型的输入变量中,综合考虑了自然环境与历史负荷因素,结合改进的布谷鸟优化的支持向量机模型(ICSSVM)进行分项预测,综合得到最后的预测结果,仿真结果表明,上述模型能得到更精确的预测结果.