灌溉用水需求预测方法初步研究

A Preliminary Study on Predicting Method for the Demand of Irrigation Water Resource

准确的分析预测灌溉用水需求及需水结构,对合理选择区域经济发展模式与发展速度至关重要。但由于问题本身的复杂性、影响因素的不确切性、特别是预测方法的局限性,使得目前需水预测误差较大。在剖析现有主要灌溉用水需求预测方法基础上,提出了需水预测模型应该考虑的基本原则及将时间序列分析的滑动平均模型、回归分析模型和人工神经网络模型有机结合构建的非线性自回归滑动平均的动态神经网络模型。

It is key to an area in choosing the proper development models of economic and the appropriate development speed that the demand and structure of irrigation water is analyzed and predicted exactly. But the complexity of the problem, the indefinite of the factors and especially the limits of the predicting method made large error between the prediction value and the reality. Based on the anatomy of the main existed predicting methods for the demand of irrigation water, this study proposes that some basic principals should be considered and a hybrid neural networks nonlinear auto-regressive moving average (NNARMAX) model has been developed, which combines the time series moving average model, auto-regression model and neural network model.