摘要
采用优化的最小二乘支持向量机方法进行轴承磨损寿命的预测。应用免疫粒子群算法对最小二乘支持向量机的建模参数进行了优化。试验样本数量通过预测得到了扩展。针对预测后的试验样本对可能服从的四种寿命分布类型进行了参数估计以及假设检验,确定了故障率分布及特征参数。试验结果表明:该方法精确度高。通过传统小样本方法与预测数据计算特征参数对比可知,预测数据计算所得的分布参数误差小,提出的预测方法合理。
A method of parameter optimized least square support vector machine was given out to predict more data based on the small sample data obtained from experiments,and the parameter＇ optimization made use of the immunity-particle swarm algorithm, so the sample＇s number of bearing＇s wear life was expanded. The parameters of four kinds of possible life distributions were estimated using the failure data from prediction, and the actual life distribution of the bearings was confirmed by distribution tests. Finally, an example was analyzed in detail for the method and the results show that the bearing failure rate is Weibull distribution. Comparision among three ways shows that the method given out herein is excellent.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2009年第11期1275-1279,共5页
China Mechanical Engineering
基金
国家863高技术研究发展计划资助项目(2006AA04Z427)
国家自然科学基金委员会与中国民用航空总局联合资助项目(60672164)
关键词
轴承
磨损寿命
最小二乘支持向量机
免疫粒子群
故障分布
bearing
wear life
least square support vector machine
immunity- particle swarm
failure distribution
