基于GA-BP神经网络的航空涡轮组件再制造工时预测  

Prediction of remanufacturing man-hours for aero-engine turbine components based on GA-BP neural network

作　　者：刘文杰[1] 余昊 陈冬梅 LIU Wenjie;YU Hao;CHEN Dongmei(College of Economics and Management,Nanjing University of Aeronautics and Astronautics,Nanjing 211106;Wuxi Mingheng Hybrid Power Technology Co.,Ltd.,Wuxi 214177;Aircraft Maintenance&Engineering Corporation,Beijing 100621)

机构地区：[1]南京航空航天大学经济与管理学院,江苏南京211106 [2]无锡明恒混合动力技术有限公司,江苏无锡214177 [3]北京飞机维修工程有限公司,北京100621

出　　处：《机械设计》2023年第8期69-75,共7页Journal of Machine Design

基　　金：国家自然科学基金资助项目(71871117);教育部人文社会科学研究一般项目(18YJA630066);航空基金(2017ZG52080)。

摘　　要：如何快速准确测定航空涡轮组件再制造工时是严重困扰我国航空发动机大修企业的一个难题。文中首次辨识和量化影响航空涡轮组件再制造工时的主要因素;之后构建一种新的再制造工时预测GA-BP神经网络。该网络采用遗传算法GA弥补了BP神经网络收敛慢且容易限于局部最优解的不足,有效提高了预测的精准度;最后,以某型民用航空发动机涡轮组件为例,验证了该预测网络的科学有效性。研究能够为航空发动机大修企业开展再制造工时预测提供高效决策支持。How to determine the remanufacturing man-hours for aero-engine turbine components accurately is one of difficult problems faced by the aero-engine overhaul enterprises.In this article,it is the first time to identify and work out the main influencing factors of the remanufacturing man-hours for aero-engine turbine components.Then,the GA-BP neural network is constructed to predict the remanufacturing man-hours.The network uses the Genetic Algorithm(GA)to compensate for the BP neural network’s slow convergence and limit of local optimal solutions;as a result,it enjoys higher accuracy in prediction.Finally,with the civil aero-engine turbine components as an example,it is verified that the prediction network is scientific and effective.This study provides efficient decision-making support for the aero-engine overhaul enterprises to predict the remanufacturing man-hours.

关 键 词：航空涡轮组件 遗传算法 BP神经网络 再制造工时预测 

分 类 号：TP307[自动化与计算机技术—计算机系统结构] V2[自动化与计算机技术—计算机科学与技术]