农田土壤湿度的人工神经网络预测诊断系统

采用人工神经网络（ＡＮＮ）方法，研制了农田土壤湿度预测，诊断分析系统，该系统根据不同的土壤湿度特点和影响因子，可自动提供时间序列拓展的人工神经网络预测模型和人工神经网络的混合预测模型进行土壤湿度的动态预测。并提供不同降水量的土壤湿水分渗透深度诊断分析，系统采用中文菜单显示的人机对话方式操作，使具有准动力学特性，多种可调参数和网络结构的ＡＮＮ方法变得简便易于操作。实际使用表明，该系统对土壤湿度的预测

Prediction and Diagnostic Analysis System of the Farmland Soil moisture through Artificial Neural Network Method

In terms of an Artificial Neural Netework(ANN),a forecast and diagnostic analysis system of the farmland soil moisture is established.A mixed model and a model involving the extension of time series of the artifivial neural network may be given in the system,respectively,based on soil humidity change and physical factors influencing its change as well.This system may be used to estimate depth of infiltrating water into soil and the change of soil moisture of different soil layers.Application results show that the system operated in interactive mode with Chinese character menu is convenient and superior on prediction and diagnostic analysis for farmland soil moisture due to ANN marked by quasi dynamical characterstic,adjusting parameter and structure.The sysytem is suitable to relieve the calamity of the agrometerological drought and to manage water resources.