摘要
土壤水分的测量对于作物的正常生长以及提高作物的经济效益具有重大意义,通过实现对土壤水分的动态预测进一步实现对作物的科学供水,对实现精准农业具有重要意义。课题组通过对现有网络预测模型的研究,发现传统的土壤含水量预测主要使用依据以往经验推导出来的预测公式进行计算,参数固定,不具备实时性的特点。基于此,课题组建立了BP动态多因素神经网络模型和RNN动态多因素土壤水分预测模型,对两种土壤水分动态预测模型进行研究,研究结果表明,RNN模型的动态多因素土壤墒情预测具有更好效果。
The measurement of soil moisture is of great significance for the normal growth of crops and improving the economic benefits of crops.Through the dynamic prediction of soil moisture,it is of great significance to further realize the scientific water supply for crops and the realization of precision agriculture.Through the research of the existing network prediction model,the research group found that the traditional prediction of soil water content mainly uses the prediction formula derived from the previous experience,with fixed parameters and no real-time characteristics.Based on this,the research group proposed the establishment of BP dynamic multi factor neural network model and RNN dynamic multi factor soil moisture prediction model,and studied the two kinds of soil moisture dynamic prediction models.The results show that RNN dynamic multi factor soil moisture prediction model has better effect.
作者
梁鑫婕
李卫东
孟凡谦
张海啸
李志伟
秦丹阳
Liang Xinjie;Li Weidong;Meng Fanqian;Zhang Haixiao;Li Zhiwei;Qin Danyang(College of information science and engineering,Henan University of Technology,Henan Zhengzhou 450001;School of marine technology,Ocean University of China,Shandong Qingdao 266000)
