根据GM(1,1)、BP神经网络、卡尔曼滤波的特点,建立以灰色关联度最大为准则的最优非负变权组合预测模型,在一定程度上抑制了误差"放大"的效应。与以误差平方和最小为准则的变权组合模型、各最优加权组合模型、各单一模型进行对比分析,结果表明,本文模型预测精度最高。
According to the characteristics of GM(1,1),BP neural network and Kalman filter,we propose an optimal non-negative variable weight combination forecasting model based on grey correlation.This new model can control the effects of the errors.The calculation results are compared with the variable weight combination model based on the error square of minimum and the optimal weighted combination model or the three single method.The results show that the new model has the highest accuracy.