基于支持向量回归的风电场短期功率预测

Short-Term Forecast of Wind Farm Power Based on Support Vector Regression Model

针对风电场的短期功率预测,提出了一种考虑风电机组运行条件的用于风电场短期功率预测的新方法.首先,利用风力发电机的监控和数据采集(SCADA)系统数据计算输出功率和运行条件之间的皮尔逊相关系数,验证了SCADA监测项目对风力发电机输出功率的具有相关性;其次,建立支持向量回归(SVR)模型来预测单个风力发电机的风力与气象、运行状态的关系,发现了考虑运行条件的模型的预测结果优于仅考虑气象信息的模型的预测结果;最后,考虑到不同空间位置的风力发电机组对风电场输出功率的贡献不同,建立了各风力发电机预测功率和风电场预测功率输出之间的回归模型.试验结果表明:所提出的风场回归模型的预测误差小于风力涡轮机所有预测功率的模型的预测误差,从而验证了该方法的有效性.

Aiming at the short-term power forecasting of wind farm,the paper proposed a new method which takes the operating conditions of wind turbines into account.Firstly,according to the analysis of the Pearson correlation coefficient between the output power and operating conditions,the specific relevance between SCADA monitoring project and output wind of wind turbines can be revealed.Then a Support Vector Regression model was built to predict the relationship among the wind power of a single wind turbine,meteorological information and the operation state of the wind turbine.The prediction results of the model which considered the operating conditions are better than those of the model considered only meteorological information.Finally,considering the difference contribution of wind turbines which lies in the different spatial positions,the regression model of each wind power generation and the wind farm＇s output power were established.The prediction error of the wind field regression model proposed in this paper is less than all the predicted power models of the wind turbine,which verified the validity of the algorithm.