基于改进PSO-LSSVM的风电场短期功率预测  

Short-Term Power Forecasting of Wind Farm Based on an Improved PSO-LSSVM

作　　者：余健明[1] 马小津[1] 倪峰[1] 王小星[1] 

机构地区：[1]西安理工大学自动化学院,陕西西安710048

出　　处：《西安理工大学学报》2013年第2期176-181,共6页Journal of Xi'an University of Technology

摘　　要：风速和风电场功率预测是风电场稳定运行及系统调度的重要保障,LSSVM在保持SVM的基础上,可以降低计算复杂性,加快求解速度,为风速及功率预测提供了一个新的研究方向。本研究将最小二乘支持向量机(LSSVM)用于风电场短期风速及风电场功率预测,提出了基于LSSVM的风电场短期风速及功率预测模型,同时建立改进粒子群模型对LSSVM进行参数优化,以内蒙古某风电场实测数据为例进行验证,实例验证表明,改进的PSO-LSSVM模型的预测效果最优。Wind speed and wind farm power forecasting are an important guarantee of stable operation and system scheduling for wind farm, while LSSVM can reduce the computation complexity, sped up so- lution speed in the foundation of SVM, provide a new research direction for the wind power forecast. This research uses LSSVM to the wind farm short-term wind speed and power forecasting, and proposes wind farm short-term wind speed and power forecast based on the LSSVM. Simultaneously, establishes the im- provement PSO model to carry on the optimization to the LSSVM parameter, and carries on the confirma- tion test by taking the Inner Mongolian some wind farm measured data as the example, the example con- firmation tests indicate that the forecast effect of improved PSO-LSSVM model is optimum.

关 键 词：最小二乘支持向量机(LSSVM) 风速 功率 预测 风电场 粒子群(PSO) 

分 类 号：TM712[电气工程—电力系统及自动化]