排刷式剥叶元件剥叶性能的BP 神经网络预测

由于剥叶性能直接影响到甘蔗收获机械的收割性能,为了客观有效地对剥叶性能进行预测,提出了BP神经网络预测方法.针对剥叶元件性能的特点,采用正交试验法构造训练样本,以保证网络具有较高的泛化能力,同时对该训练样本建立了回归分析模型,以检验BP网络模型的输出精度.在此基础上,利用已建立的神经网络预测模型对影响剥叶性能的各因素取值的不同组合进行仿真分析,以确定各因素取值的最优组合.结果表明,BP神经网络的预测模型比回归模型具有更高的输出精度,进行剥叶元件的性能预测与优化是可行且有效的.

Performance Forecasting Based on BP Neural Network for Cleaning Element in Brush Shape

The cleaning performance affects the harvest performance of the whole-stalk sugarcane harvester.In order to forecast the cleaning performance effectively,the performance forecast model based on BP neural network is presented.According to the characteristics of cleaning performance,the training samples are made up of the orthogonal experimental data,which are also used to build the regression analysis model to examine the output precision and to ensure higher generalization of BP neural network.And then the trained BP neural network is used to forecast and analyze on the different value combinations of the factors influencing the cleaning performance.Consequently,the optimal combination is determined.The results show that the output precision of BP neural network is higher than that of the regression analysis model,and using the BP neural network to forecast the cleaning performance is practicable and effective.