风电机组齿轮箱温度预测中输入变量的选择直接影响预测的准确性 ,本文将正交最小二乘法和相关系数法用于齿轮箱输入变量的选取 ,采用BP神经网络 ,对比预测了齿轮箱温度 .研究结果表明 :选取的输入变量用于温度预测的均方误差、均方根误差和平均绝对百分比误差均小于相关系数法 ,验证了正交最小二乘法在齿轮箱温度预测输入变量选择上的有效性 .%For the temperature prediction of wind turbine gearbox ,selecting the input variable is one of the most important procedure ,which is connected with the results of prediction directly .This article introduces orthogonal least squares .Using orthogonal least squares and correlation coefficient to select input variable of gearbox ,and applying the results in BP neural network to predict the temperature of gearbox .The results show that using orthogonal least squares has a smaller MSE RMSE and MAPE than that of correlation coefficient ,thus its effectiveness is demonstrated .
展开▼