基于遗传算法的人工神经网络在航空发动机磨损预测中的应用  

Application of BP Neural Network Based on GA to Predicting Wear of Aeroengine

作　　者：姜旭峰[1] 费逸伟[1] 王惠[1] 钟新辉[1] 

机构地区：[1]徐州空军学院,江苏徐州221006

出　　处：《润滑与密封》2007年第2期168-170,188,共4页Lubrication Engineering

摘　　要：提出了一种将遗传算法和BP算法相结合的学习算法来训练BP神经网络,实现网络结构的优化,并用优化后的BP人工神经网络建立了航空发动机的磨损预测模型。将该模型预测结果与BP算法和多元线性回归法的预测结果进行了比较。检验结果表明:基于遗传算法的BP神经网络优于BP算法及多元线性回归法,具有良好的预测效果。A prediction model based on the GA（ genetic algorithm） and BP（ back propagation） network by using GA to train BP neural network and optimize network structure was presented. The wear of aeroengine was estimated by the optimized network. The resuh of this approach was compared with that of the simple BP network as well as Poly-regression model. The results indicate that BP neural network based on GA is superior to the simple BP network and the Poly-regression model.

关 键 词：遗传算法 BP神经网络 航空发动机 磨损 预测 

分 类 号：X131.1[环境科学与工程—环境科学]