【摘要】 风速、风向的随机性导致风电场功率具有很大波动，对风电功率的精确预测在提高电网运行能力，增强电网接收风电能力和适时安排风电场检修计划等方面提供重要依据。提出了一种粒子群优化核极限学习机(PSO-KELM)算法；对风电机组SCADA数据进行预处理，补充和纠正异常数据；针对极限学习机的复共线性问题提出基于核函数的改进极限学习机，避免了极限学习机输出结果的随机性；采用粒子群优化算法对KELM核函数中的惩罚因子和径向参数值进行优化，建立了基于粒子群优化的极限学习机模型。与BPNN与RBFNN等其他方法相比，真实风电场数据验证了该模型的预测精度。更多还原

【Abstract】 Randomness of wind speed and wind direction leads to large fluctuations in wind fleet capacity.Accurate forecasting of wind energy can provide an important basis for improving the power grid performance, enhancing the power grid’s ability to receive wind power and timely arranging the maintenance plan of wind farm.This document provides the particle Legion optimization kernel restricted learning machine(PSO-KELM) algorithm.Preprocess SCADA data of wind turbine generator, supplement and correct abnormal data; Aiming at the multicollinearity problem of extreme learning machine, A foreign learning machine is proposed on the basis of nuclear operation to prevent accidents from driving out a foreign learning machine.The optimization algorithm of particle emissions(PSO) is used to optimize the penile factor and the radiation parameters of the KELM nucleus and generate an external learning model based on optimization of particle emissions.Go on.Compared to other ways like BPNN and RBFNN,this model is confirmed by real wind farm data.更多还原