基于L-M优化算法的BP神经网络的作物需水量预测模型

应用L-M优化算法BP神经网络，通过多维气象数据(太阳辐射、空气温度、湿度)与作物需水量的相关分析，来确定网络的拓扑结构，建立作物需水量的人工神经网络模型。用美国田纳西州大学高原实验室所测的100 d气象数据为输入、作物需水量为输出来训练建立好的BP神经网络，仿真表明该神经网络能很好地解决需水量多影响因素之间的不确定性和非线性，模型的预测精度较高，同时通过一组非样本天气环境参数和作物需水量来验证该神经网络，也得到了较好的预测结果，能够满足灌溉的精度要求。

Model for predicting crop water requirements by using L-M optimization algorithm BP neural network

In this paper, a network topological structure was determined and an artificial neural network model for predicting crop water requirements was established by using L-M optimization algorithm BP neural network and correlation analysis between multi-dimension climate data and crop water requirements. The BP neural network was trained by experimental meteorological data of 100 days measured in Tennessee Plateau Experiment Station. The simulated results showed that the BP neural network can solve the uncertainty and non-linearity of multi climate factors, and the prediction precision of the model is high. At the same time, the neural network prediction precision was tested by a group of non-specimen meteorological data and crop water requirements. The tested results were good enough to meet the requirements of irrigation precision.