基于振动试验和神经网络的航空导线寿命预测方法研究  

Research on Aviation Wire Life Prediction Method based on Vibration Test and Neural Network

作　　者：刘科文[1] 乐洪博 王宇宁 谢大叶 LIU Kewen;LE Hongbo;WANG Yuning;XIE Daye(Engineering Technology Center,AVIC Shenyang Aircraft Industry(Group)Co.,Ltd.,Shenyang 110850,China;First Military Representative Office of Ministry of Air Force Equipment in Shenyang,Shenyang 110850,China)

机构地区：[1]航空工业沈阳飞机工业(集团)有限公司工程技术中心,辽宁沈阳110850 [2]空装驻沈阳局驻沈阳地区第一军事代表室,辽宁沈阳110850

出　　处：《新技术新工艺》2024年第9期70-74,共5页New Technology & New Process

摘　　要：面向航空导线传统预测手段是基于经验为主的预测,不仅误差高,而且经验性过强,并没有形成一定的定性分析,对于其寿命预测的准确性缺乏科学性。针对这种缺陷提出了一种创新方法,首先对飞机导线人工创造损伤,然后进行恒定量级随机振动磨损试验,进而运用不同代理模型的建模与参数估计方法并采用有限差分法对飞机导线服役期间全周期寿命和可靠寿命进行预测,最终发现了导线磨损失效的规律,达到了预测效果,解决了寿命预测缺乏科学性的问题,具备了一定的定性分析能力,形成的寿命预测方法对于航空企业具有指导意义。The traditional prediction methods for aviation wires were mainly based on experience,which not only had high errors,but also had strong experience and did not form a certain qualitative analysis,so the accuracy of their life prediction lacked scientificity.In response to this deficiency,we proposed an innovative method.Firstly,artificial damage was created to the aircraft wires,followed by constant magnitude random vibration wear tests.Then,modeling and parameter estimation methods using different surrogate models were applied,and the finite difference method was used to predict the full cycle life and reliable life of the aircraft wires during service.Finally,the law of wire wear failure was discovered,and achieved the prediction effect and solved the problem of lack of scientificity in life prediction.The formed life prediction method had certain qualitative analysis ability and guiding significance for aviation enterprises.

关 键 词：导线 微动磨损 振动试验 神经网络 代理模型 峰值载荷 

分 类 号：TG659[金属学及工艺—金属切削加工及机床] TP183[自动化与计算机技术—控制理论与控制工程] TH162[自动化与计算机技术—控制科学与工程] TH39[机械工程—机械制造及自动化]