安徽工程大学学报
安徽工程大學學報
안휘공정대학학보
JOURNAL OF ANHUI UNIVERSITY OF TECHNOLOGY AND SCIENCE(NATURAL SCIENCE)
2012年
3期
65-68
,共4页
风速预测%小波神经网络%置信区间%预测精度
風速預測%小波神經網絡%置信區間%預測精度
풍속예측%소파신경망락%치신구간%예측정도
Wind speed forecasting%Wavelet neural network%Confidence interval%Forecast accuracy
小波神经网络是在小波变换理论和人工神经网络的基础上建立的一种新型网络模型，综合了两者的优点，克服了BP神经网络易陷入局部极小点和训练速度慢的缺点．本文建立了小波神经网络模型，采用最陡梯度下降法训练网络，将该网络用于对风电场小时风速的预测，并对预测置信区间进行计算．预测结果表明小波神经网络在训练速度和预测精度方面均优于BP神经网络．
小波神經網絡是在小波變換理論和人工神經網絡的基礎上建立的一種新型網絡模型，綜閤瞭兩者的優點，剋服瞭BP神經網絡易陷入跼部極小點和訓練速度慢的缺點．本文建立瞭小波神經網絡模型，採用最陡梯度下降法訓練網絡，將該網絡用于對風電場小時風速的預測，併對預測置信區間進行計算．預測結果錶明小波神經網絡在訓練速度和預測精度方麵均優于BP神經網絡．
소파신경망락시재소파변환이론화인공신경망락적기출상건립적일충신형망락모형，종합료량자적우점，극복료BP신경망락역함입국부겁소점화훈련속도만적결점．본문건립료소파신경망락모형，채용최두제도하강법훈련망락，장해망락용우대풍전장소시풍속적예측，병대예측치신구간진행계산．예측결과표명소파신경망락재훈련속도화예측정도방면균우우BP신경망락．
Abstract.The wavelet neural network (WNN) is a new model based on wavelet theory and artificial neural net- work. It combines the advantages of those two theories and overcomes the shortcomings that BP neural network is easy to get into a local minimum point and reduce the train speed. The WNN model is established and the steep gradient descent method is used to train network. The WNN is applied to predicting hourly wind speed of wind power plant and confidence interval is calculated also. The forecast results indicate that the WNN proposed in the article is better than BP network in train speed and forecast accuracy.
