基于模糊最小二乘支持向量机的系统边际电价预测方法

System Marginal Price Forecasting Using Fuzzy Least Square Support Vector Machine

提出了基于模糊最小二乘支持向量机的系统边际电价(system marginal price,SMP)预测方法。为了减少样本数据中孤立点对回归性能的影响,将模糊隶属度的概念引入到最小二乘支持向量机中的同时,采用网格搜索和交叉验证的方法寻找最佳参数组合,使系统边际电价算法性能达到最佳。以美国加州电力市场的实际数据作计算实例,分别采用标准三层BP神经网络和模糊最小二乘支持向量机进行系统边际电价预测,结果表明基于模糊最小二乘支持向量机的系统边际电价预测的方法有效提高了预测精度。

A method based on fuzzy least square support vector machine （FLS-SVM） is proposed for system marginal price （SMP） forecast. The concept of fuzzy membership is introduced into FLS-SVM to reduce the effects of sample data outliers on regression performance. Meanwhile, grid search and cross validation are adopted to search the best parameters and enable optimal performance of SMP algorithm. Then taking the data of California electricity market for calculation example, the SMP forecasting is performed by FLS-SVM and the standard 3-layer BP neural network. The results show that the proposed method effectively increases the forecasting precision.