Abstract: To accurately evaluate the performance degradation trend of rolling bearings,a Type-ⅡGeneralized Logistic Distribution(Ⅱ-GLD)position calibration model was proposed to evaluate the reliability of rolling bearings performance degradation.CombiningⅡ-GLD with the position calibration model,the mean square error between the degradation of model at each moment and the test samples were calculated as the fitness function.Then,the step factor of Firefly Algorithm(FA)was improved adaptively,and the disturbance factor was introduced to improve the searching ability of the algorithm.Finally,the parameters of Ⅱ-GLD position calibration model were optimized by using improved FA.Kernel Principal Component Analysis(KPCA)was used to filter reconstruction features with cumulative contribution rate greater than 95% and input them into the optimal evaluation model for experimental verification.The test results showed that the relative errors of the mean and variance between the improved FA-optimized Ⅱ-GLD position calibration model and the original data were below 2% and 8%,respectively,and its evaluating performance was better than Weibull risk scale model and normal distribution.At the same time,the envelope spectrum analysis of the bearing reliability was carried out,and the results were highly consistent with the actual situation,which fully verified the accuracy and validity of the model.

Key words: rolling bearing, firefly algorithm, Type-Ⅱgeneralized logistic distribution position calibration model, performance degradation evaluation

摘要： 为准确评估滚动轴承的性能退化趋势,提出一种Ⅱ型广义logistic分布(Ⅱ-GLD)位置刻度模型进行滚动轴承性能退化可靠性评估。将Ⅱ-GLD与位置刻度模型结合,计算模型各时刻退化量与试验样本间的均方误差作为适应度函数;对萤火虫算法(FA)的步长因子进行自适应改进,同时引入扰动因子提高算法搜索能力;利用改进FA优化Ⅱ-GLD位置刻度模型的参数。核主成分分析法筛选累积贡献率大于95%的重构特征,输入到最优评估模型进行试验验证。结果表明,改进FA优化的Ⅱ-GLD位置刻度模型的均值、方差与实际数据间的相对误差分别低于2%和8%,相比威布尔模型和正态分布,其评估性能更佳。同时对实例轴承的可靠性进行包络谱分析,结果与实际状况高度吻合,充分验证了模型的准确性和有效性。

关键词: 滚动轴承, 萤火虫算法, Ⅱ-GLD位置刻度模型, 性能退化评估