基于BP神经网络的排种器充种性能预测

Prediction for Performance of Seed-metering Process Based on BP Neural Network

在分析丸粒化玉米种子精密排种器工作原理的基础上,运用Matlab神经网络工具箱建立了排种合格指数A的BP神经网络预测模型,选取排种轴转速n、型孔直径D和排种盘的锥角Ф作为试验因素进行排种性能试验,得到81组合格指数的试验结果。选取其中的60组进行样本训练,并对剩下的21组结果对训练好的网络进行预测,n,D,Ф为输入层,A为输出层,网络结构为3-11-1的3层网络。预测结果表明,预测值与试验值误差较小,为排种器的优化设计及参数的选择提供了依据。

According to the analysis of the working principle of the precision seed-metering device based on the pelleted seeds,the BP neural network prediction model filled at qualification index was established using the Matlab neural network toolbox.The rotary speed n,the diameter of the cell D,the cone angle of the moving disc were selected as the test factors,the test was carried out 81 groups to determine the qualification index and 60 groups were selected from the test as training samples.The remaining 21 groups were selected to simulate and predict the trained neural network.N,D and Ф were set as the network＇s input layers,A was set as the network＇s output layer.The network structure was the 3-11-1 type three-layer network.Predicted results showed that predicted values and experimental values had little error and this provided basis for the optimum design of the seed metering.