基于灰色关联分析与BP神经网络的农业灌溉预测

Agricultural Irrigation Forecasting Based on Grey Relational Analysis and BP Neural Network

针对农业灌溉用水量预测存在的复杂性、不确定性和非线性等问题,提出一种基于灰色关联度分析与BP神经网络的灌溉需水量预测,首先,采用灰色关联度分析法,选取降水量、蒸发量、平均气温、日照时间、灌溉面积作为BP神经网络的输入因子;然后,根据各影响因子与灌溉用水量的对应关系,对模型训练;最后,将训练好的模型用于2007-2017年灌溉需水量预测中.结果表明,灰色关联-BP神经网络模型预测相对误差在1.81%~5.48%以内,可为农田灌溉预测提供科学依据.

In view of the complexity,uncertainty and nonlinearity of agricultural irrigation water consumption forecasting,the irrigation forecasting based on grey relational analysis and BP neural network is proposed.Firstly,precipitation,evaporation,average temperature,sunshine time and atmospheric pressure are selected as the input variables of BP neural network based gray correlation analysis method,then,the model is trained based on the corresponding relationship between each impact factor and irrigation water use,finally,the trained model is used in the forecast of irrigation water use between 2007 and 2017.The results show that the relative error of prediction by the grey relation analysis-BP neural network model is between 1.81% and 5.48%,which can provide scientific guidance for farmland irrigation.