小样本下基于GM-SVR方法的航空保障特种装置可靠性评估  

Reliability Evaluation of Aviation Support Special Equipment Based on GM-SVR Method in Small Samples

作　　者：刘洋 宋庭新[1] 谈太阳 LIU Yang;SONG Tingxin;TAN Taiyang(School of Mechanical Engineering,Hubei University of Technology,Wuhan Hubei 430068,China)

机构地区：[1]湖北工业大学机械工程学院,湖北武汉430068

出　　处：《机床与液压》2024年第9期201-208,共8页Machine Tool & Hydraulics

基　　金：国家重点研发计划资助项目(2018YFF0214705)。

摘　　要：航空保障特种装置的可靠性评估面临失效样本数据偏少的问题。为更加准确评估小样本情况下的可靠性,采用改进的平均秩次法计算样本的经验分布函数,在此基础上提出一种基于GM-SVR的三参数威布尔分布估计方法。该方法利用灰色模型(GM)在位置参数上的优势和支持向量回归(SVR)在小样本下的良好效果,对威布尔分布的3个参量进行精确估计。采用MATLAB工具箱对模型进行仿真实验,结果表明:GM-SVR方法在故障数据样本较少情况下,参数估计精度高于传统的概率统计方法。以喷气偏流板为例,计算了威布尔分布的形状参数、尺度参数和位置参数,得到相关系数为0.9953,均方根误差为0.038,证明了在小样本情况下GM-SVR方法的拟合精度良好,可对航空保障特种装置的可靠性进行准确评估。The reliability evaluation of aviation support special equipment is faced with the problem of insufficient failure sample data.In order to evaluate the reliability of small samples more accurately,the improved average rank method was used to calculate the empirical distribution function of samples,and a three-parameter Weibull distribution estimation method was proposed based on GM-SVR.In this method,the advantages of grey model(GM)in location parameters and the good effect of support vector regression(SVR)in small samples were used to accurately estimate the three parameters of Weibull distribution.The simulation experiment for the model was carried out using MATLAB toolbox.The results show that the GM-SVR method has higher parameter estimation accuracy than the traditional probabilistic statistical method when the fault data samples are small.Taking the jet deflector as an example,the shape parameter,scale parameter and position parameter of Weibull distribution were calculated,the correlation coefficient was 0.9953,and the root mean square error was 0.038.It proves that the GM-SVR method has good fitting accuracy in the case of small samples,and can be used to accurately evaluate the reliability of special aviation support devices.

关 键 词：灰色模型 支持向量回归 可靠性分析 航空保障装置 

分 类 号：TH17[机械工程—机械制造及自动化]