基于DE-GWO-LSSVM模型的用电负荷预测

Research on Power Load Forecasting Based on DE-GWO-LSSVM Model

为了提高负荷预测的拟合精度,提出一种基于优化灰狼算法的最小二乘支持向量机负荷预测模型,针对标准灰狼算法精度低、收敛速度慢、易陷入局部最优的缺点,采用差分算法优化标准灰狼算法。利用改进的灰狼算法优化最小二乘支持向量机的两个主要参数,建立功率负荷预测研究模型。通过实例分析获得负荷预测结果,利用三种评价指标对比了四种算法模型。实验表明,改进灰狼算法优化最小二乘支持向量机的改进评价指标数值较低,拟合曲线精度更高。

In order to improve the fitting accuracy of load forecasting,a least squares support vector machine load forecasting model based on optimized grey wolf algorithm is proposed.The proposed model is aimed at the shortcomings of standard gray wolf algorithm,such as low precision,slow convergence speed and easy falling into Locally optimal conditions.The differential algorithm is used to optimize the standard gray wolf algorithm.The improved grey wolf algorithm is then applied to optimize the two main parameters of the least squares support vector machine,by which the power load prediction research model is established.In this paper,load forecasting results are obtained by case analysis.Three kinds of evaluation models are used to compare the four algorithm models.Experiment results show that the improved evaluation index of the modified gray-horse algorithm optimized least squares support vector machine is lower,and the accuracy of the fitting curve is higher.