基于遗传神经网络的风电场输出功率短期预测  

Short-Term Forecast of Wind Farm Output Power Based on Genetic Algorithm-Neural Network

作　　者：罗毅[1] 刘峰[1] 

机构地区：[1]华北电力大学控制与计算机工程学院,北京102206

出　　处：《华东电力》2012年第7期1205-1208,共4页East China Electric Power

基　　金：国家自然科学基金项目(60974051)~~

摘　　要：针对风电场短期输出功率的时变性和非线性特点,提出了基于遗传神经网络的短期风电功率预测方法。结合BP网络对非线性问题良好的逼近能力和遗传算法优良的全局寻优能力,遗传算法有效解决了BP神经网络全局搜索能力差、易陷入局部极小值的问题。与BP神经网络模型相比,遗传神经网络模型预测精度有所提高,为风电场输出功率短期预测提供了一种有效的方法。Considering the time variation and nonlinearity of short-term wind farm output power, this paper proposes a forecast method based on genetic algorithm and neural network （GANN）. GANN model incorporates the superiority of BP neural network in solving nonlinear problem and the ability of genetic algorithm （GA） in global optimization, which can effectively avoid the defects of BP neural network in the deficiency of global searching and vulnerability to fall into local minima. Compared with BP neural network model, the GANN model can improve the forecast accuracy, proved an effective way to forecast short-term wind farm output power.

关 键 词：风电场 短期风电功率 预测 BP神经网络 遗传算法 

分 类 号：TM614[电气工程—电力系统及自动化]