基于GA优化的加权LSSVM短期风速预测

Short-term wind speed forecasting study based on GA optimized weighted LSSVM model

精确的风电场风速预测可以提高风力发电在电力市场中的竞争力,还可提高电力系统的稳定性。为了降低短期风速预测的误差,提出了基于遗传算法(GA)优化加权最小二乘支持向量机(WLSSVM)的短期风速预测模型。该模型以风场实测风速数据作为模型的输入向量,根据遗传算法对加权最小二乘支持向量机的惩罚参数和核函数参数寻找最优解,以此建立起参数最优的风速预测模型。该模型用于研究某风电场同一季节连续的300个(采样间隔1h)历史风速数据,取前240个数据为训练集,后60个数据为预测集,预测结果的平均绝对百分比误差仅为11.88%。与只采用最小二乘支持向量机(LSSVM)进行预测的模型对比,该模型预测精度较高。

The accurate wind farm wind speed prediction can improve the competitiveness of wind power in the electricity market and improve the stability of the power system. In order to reduce the error of short-term wind speed prediction, a short-term wind speed forecasting method based on genetic algorithm (GA) optimized weighted least square support vector machine (WLSSVM) model is proposed. In this model, genetic algorithm is employed to optimize the penalty factor and kernel parameter of weighted support vector machines, in which, the wind farm of actual measurement vale wind speed is taken as input vector, and then the prediction model of wind with optimal parameters is established. The model is used to study the 300 continuous historical wind speed data (sampling interval 1h) of a wind farm in the same season. The first 240 data are the training set, and the last 60 data are the prediction set. The average absolute percentage error of the prediction result is only 11.88%. The accuracy of GA-MLSSVM model is much higher than that of LSSVM.