电力系统短期负荷预测建模仿真研究

Pretreatment of Short-Term Load Forecasting Based on K-Means Clustering Algorithm

精确的电力系统负荷预测能够使电网安全稳定的运行。传统的电力系统负荷预测只注重预测方法的研究,诸如基于人工神经网络的方法、支持向量机的方法等,很少关注数据的预处理。为了提高短期负荷预测的精度,采用双向比较法对浙江省某市的实际电力负荷历史数据进行预处理,并用K-means算法进行聚类分析,使具有相同特征的数据属性归为一类,以此降低原始数据维数。利用LSSVM算法进行负荷预测,从而得到准确的预测结果。仿真结果表明,经过聚类的LSSVM模型预测结果的平均相对误差和最大相对误差,远低于传统模型。充分说明了双向比较法和K-means算法相结合的短期负荷预测方法不仅有更高的预测精度还使预测误差更加的平稳。

Accurate load forecasting of power system can make the power grid safe and stable operation. Traditional power system load forecasting only pays attention to the forecast method of research,such as the method based on ANN,SVM method and so on,seldom focuses on data preprocessing. In order to improve the accuracy of short-term load forecasting,we used two-way comparison method to preprocess the actual power load historical data in Zhejiang Province,and then used the K-means clustering algorithm,to make the data attributes to have the same characteristics and to reduce raw data dimension,thereby reducing the original data dimension. LSSVM was used to load forecasting,and accurate prediction results were achieved. The simulation results show that after clustering,the predicted average relative errors and maximum errors based on LSSVM model are far lower than traditional models,which fully illustrates that the bidirectional comparison method and K-means algorithm with the combination of short-term load forecasting method not only has a higher prediction precision but also makes prediction error more smoothly.