灌溉用水量的并联型灰色神经网络预测

Prediction of irrigation water use using parallel gray neural network

摘要: 该文提出了把人工神经网络和灰色预测方法结合成并联型灰色神经网络预测方法，用这种方法来预测灌溉用水量，并以预测方法有效度为优化指标求解组合模型加权系数。结果显示，灰色神经网络预测方法的平均误差为2.67%，明显低于单一的灰色预测方法和神经网络预测方法的平均误差，可以将这种组合方法应用于中长期灌溉用水量预测。

Abstract: The paper put forward a forecast method named parallel gray neural network (PGNN), which was combined with neural network and the gray forecast method. The PGNN was adopted to forecast irrigation water use and the forecast method availability degree was used as the optimization index to calculate the weighted coefficient of the combination model. The results showed that the average error of PGNN was 2.67% and it was obviously lower than that of unitary gray and neural network forecast method. PGNN can be applied to forecast middle-long term irrigation water use.