基于最小二乘支持向量机的短期负荷预测

针对当今人工智能短期负荷预测方法存在的缺陷,提出了一种最小二乘支持向量机(LS-SVM)短期负荷预测方法,即建立最小二乘支持向量机(LS-SVM)回归模型。在选取该模型训练样本时,为了提高预测精度,采用灰色关联投影法来选取相似日。同时,针对标准粒子群优化算法易陷入局部最优的缺点,提出自适应变异粒子群优化算法来选择最小二乘向量机的参数,从而提高了负荷预测精度,避免了对模型参数的盲目选择。仿真结果分析表明,该方法有效、可行。

Short- term load forecasting based on least square support vector machine

Aiming at the flaw of the existing artificial intelligent short - term load forecasting, this paper proposes a new method based on least square support vector machine ： the regression model of least square support vector ma- chine （ LS - SVM）. When the training sample of the model is selected, grey relation projection method is adopted to select similar day in order to enhance forecasting accuracy. Besides, according to the disadvantage of the stand- ard particle swarm optimization algorithm which easily falls into local optimum, the paper also proposes to select pa- rameters of least square support vector machine by particle swarm optimization with adaptive mutation, which en- hances the accuracy of load forecasting and prevents the blind choice of model parameters. Practical simulation re- sult proves the efficiency and feasibility of this method.