基于ANN的轴流脱粒与分离装置性能指标预测

影响水稻轴流脱粒与分离装置性能指标的因素很多,并且它们之间存在着复杂的非线性关系,用传统方法很难对其进行准确预测。神经网络算法简单、学习收敛速度快、具有线性、非线性逼近精度高等特性。本文以正交旋转组合实验获得的数据作为样本,对人工神经网络(ANN)模型进行训练学习,利用训练后所得到的模型,对性能指标进行了预测。结果表明,网络预测值与实测值之间具有很高的相关性和精确度,为机械性能指标研究提供了一定的理论辅助手段。

Performance Indexes Prediction of Axial Threshing and Separating Mechanism Based on ANN

Traditional methods are hard to accurately predict performance indexes of grain axial threshing and separating mechanism due to a lot of factors could influence the prediction and a complicated and nonlinear relationship exists among these factors. Neural network is a simple algorithm with characteristics of easy to learn, high speed in learning process and excellent in the linear or nonlinear accurate approximation which is suitable for modeling the grain axial threshing and separating mechanism performance index. Using data from orthogonal test as sample, study on the artificial neutral network was carried out. According to the trained modes, performance indexes of the ingredients related to orthogonal test were analyzed and predicted. The results showed that the prediction precision and correlation between the predicted ANNs and measured values are considerably high. This could offer theoretical elements for prediction of mechanism performance index.