非充分灌溉青贮玉米土壤墒情预报的人工神经网络模型  

Artificial Neural Network Model for Soil Moisture Forecast in Inadequate Irrigation of Maize Harvested Green Field

作　　者：郑和祥[1] 史海滨[1] 柴建华 傅卫平 

机构地区：[1]内蒙古农业大学水利与土木建筑工程学院,内蒙古呼和浩特010018 [2]内蒙古水利科学研究院,内蒙古呼和浩特010010

出　　处：《灌溉排水学报》2006年第6期53-56,共4页Journal of Irrigation and Drainage

基　　金：内蒙古自治区科技支撑项目

摘　　要：土壤墒情预报是农田适时适量灌溉与科学管理的基础,田间土壤墒情的变化受降水、灌溉、植株蒸腾、土壤蒸发、根系层下边界水分通量及外界气象因素的影响,关系比较复杂。利用内蒙古锡林浩特市典型草原区的青贮玉米土壤水分试验资料,建立了土壤墒情预报的BP神经网络模型,并利用部分实测资料对网络进行检验,取得了较好的效果。结果表明BP神经网络模型可以对区域土壤水分进行动态预测,方法简便可行。Irrigation and scientific manage water are based on soil moisture forecast. The change of the soil moisture in the field is quite complex and mainly effected by rainfall, irrigation, transpiration, evaporation, the flux water of down boundary in root and weather factors. Based on the soil moisture observation data for maize harvested green in typical grassland in Xilinhaote of Inner Mongolia, a Back Propagation （BP） network model for soil moisture forecast is established. The network is proved by observation data and the result is well. The predicted soil moisture fairly well agrees with the observation data and the means is convenient and feasible.

关 键 词：非充分灌溉 青贮玉米 墒情预报 BP神经网络模型 

分 类 号：S513[农业科学—作物学] S548