基于APSO-LSSVM的航空发动机轴承故障诊断及寿命预测

Fault Diagnosis and Life Prediction of Aeroengine Bearings Based on APSO-LSSVM

航空发动机轴承在高速、高温、高载荷等极端工况下易发生机械故障,为了提前预警,提出了一种基于自适应粒子群优化(Adaptive Particle Swarm Optimization,APSO)算法的最小二乘支持向量机(APSO Least Squares Support Vector Machine,APSO-LSSVM)对滑油系统中轴承磨屑进行在线监测的故障诊断及寿命预测。通过主成分分析法(Principal Components Analysis,PCA)对滑油磨屑信息进行降维处理,构建特征向量,并将特征向量输入APSO-LSSVM模型,对轴承故障状态进行分类并对轴承剩余寿命进行预测。结果表明:使用PCA可以保留数据样本99.9%的信息,同时还能极大地降低数据维度;与遗传算法(Genetic Algorithm,GA)、灰狼优化(Grey Wolf Optimization,GWO)算法、粒子群优化(Particle Swarm Optimization,PSO)算法的支持向量机相比,所提算法因采用了自适应调节粒子移动步幅,在进行轴承状态分类时准确率更高,分类正确率可达95.56%,同时在进行轴承剩余寿命预测时具有较好的准确度和泛化性。

To predict the structural life time of bearings in aeroengines under extreme operating conditions such as high speed,high temperature,and high load,a fault diagnosis and lifetime prediction method utilizing adaptive particle swarm optimization and least squares support vector machine (APSO-LSSVM) for online monitoring of bearing debris in lubricating oil systems is proposed.To reduce the dimensionality of oil debris information,principal components analysis (PCA) is incorporated into the algorithm and feature vectors are constructed,which are subsequently input into the APSO-LSSVM model for classifying bearing fault states and predicting the remaining bearing life.The effectiveness of PCA in preserving a high percentage of information from data samples while significantly decreasing their dimensions is demonstrated.Compared with support vector machines optimized through genetic algorithm(GA),grey wolf optimization(GWO) algorithm,and particle swarm optimization(PSO) algorithm,the present algorithm with adaptive adjustment of particle moving step implanted exhibits better accuracy and generalizability,with a classification accuracy of 95.56% in predicting the remaining life of bearings.