发动机活塞性能预测的BP网络方法  

Forecasting of Piston Performance Based on BP Neural Networks

作　　者：马永昌[1] 代荣[1] 张建军[1] 

机构地区：[1]西南大学工程技术学院,重庆400716

出　　处：《机械科学与技术》2011年第9期1451-1455,共5页Mechanical Science and Technology for Aerospace Engineering

基　　金：国家自然科学基金项目(150375161);重庆市科技攻关项目(7823-10)资助

摘　　要：为了客观有效地对活塞性能进行预测,提出了BP神经网络预测方法。针对活塞元件性能的特点,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,同时用检验样本来比较BP(back propagation)网络模型的输出结果。结果表明:BP神经网络的预测模型具有较高的输出精度,进行发动机活塞元件的性能预测是可行且有效的。In order to forecast the piston performance effectively,the performance forecast model based on BP（Back Propagation） neural networks is presented.According to the characteristics of piston performance,the training samples are taken from the orthogonal experimental data which are also used to ensure higher generalization of BP neutral networks.Then,the trained BP neural network is used to forecast test examples.At the same time,the value of test samples is compared with the output of BP network model results.The results show that the output precision of BP neural network is high,and using the BP neural network to forecast the piston performance is practicable and effective.

关 键 词：发动机 活塞 BP神经网络 预测 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TP302.7[自动化与计算机技术—控制科学与工程]