陕西省耕种收综合机械化水平预测研究

Research on Prediction of the Comprehensive Level of Mechnical Farming,Sowing and Harvesting in Shanxi Province

研究采用BP网络改进算法、RBF网络和遗传神经网络,构建了机耕、机播与机收水平预测模型。应用所构建的模型对陕西省机耕、机播、机收水平进行预测,着重对3种预测模型进行比较和分析,为有关部门制定政策、方针和路线提供科学依据。用测试样本对预测精度进行检验,3种机耕水平预测模型(基于BP改进算法的BP模型、GA-BP模型和RBF模型)的平均相对误差都在10%以内;3种机播水平预测模型的平均相对误差分别为5.17%,9.0%和13.86%;3种机收水平预测模型的平均相对误差分别为12.11%,7.22%和16.75%。对于机耕水平预测问题,综合考虑预测精度、构建网络的复杂度和稳定性,采用RBF模型较好;机播水平预测采用预测精度最高的基于改进BP算法的BP模型;机收水平预测可采用GA-BP模型。

Research adopted the improved BP network algorithm,RBF network and genetic neural network to construct predicting models which have been used to predict the level of mechanized farming,the level of mechanical sowing,and the level of mechanical harvesting in Shaanxi Province,three predicting models was compared and analyzed in this paper.The research provides the scientific basis for the government to make related policies,guidelines,planning.Test samples were used in these models to examine Prediction Accuracy,Experimental results show that the relative average error of there models of the level of mechanized farming are all within 10%,the relative average error of there models of the level of mechanical sowing are 5.17%,9.0% and 13.86% respectively,the relative average error of there models of the level of mechanical harvesting are 12.11%,7.22% and 16.75% respectively.For the level of mechanized farming prediction,considered according to predicting precision,complexity of constructing models and stability,RBF model is better;the improved BP-algorithm-based BP model shoule be used for the level of mechanical sowing prediction,and genetic artificial neural network should be used for the level of mechanical harvesting prediction.