| 摘 要: 针对滚动轴承故障诊断中处理高维非线性特征数据的难题,提出了一种基于核主元分析(Kernel Principal Component Analysis,KPCA)和改进的白鲸优化算法(Improved Beluga Whale Optimizer,IBWO)优化支持向量机(Support Vector Machine,SVM)超参数的方法,即 KPCA-IBWO-SVM 模型。通过引入折射反向学习和旋风觅食策略,显著提升了IBWO的收敛速度和全局搜索能力。首先,利用 KPCA提取原始数据中的非线性主元特征;其次,通过SVM 模型完成故障诊断。实验结果表明,IBWO 算法相较于灰狼优化算法(GWO)、鲸鱼优化算法(WOA)、麻雀搜索算法(SSA)及原始白鲸优化算法(BWO)等具有明显优势,KPCA-IBWO-SVM 模型的平均诊断准确率达到95.86%,比 KPCA-BWO-SVM 模型提升了6.54%,充分验证了所提方法的有效性和应用价值。 |
| 关键词: 改进的白鲸优化算法;支持向量机;故障诊断;核主成分分析;滚动轴承 |
|
中图分类号: TP277;TH133.33
文献标识码: A
|
|
| Research on Rolling Bearing Fault Diagnosis Based on KPCA and IBWO-Optimized SVM |
|
WANG Jie, LIU Tianlun, QIU Yiyang
|
(School of Automotive and Traffic Engineering, Wuhan University of Science and Technology, Wuhan 430065, China)
wangjie1980@wust.edu.cn; 1941851860@qq.com; 645753223@qq.com
|
| Abstract: To address the challenge of processing high-dimensional nonlinear feature data in rolling bearing fault diagnosis, this study proposes a method combining Kernel Principal Component Analysis (KPCA) and an Improved Beluga Whale Optimizer (IBWO) to optimize the hyperparameters of a Support Vector Machine (SVM), referred to as the KPCA-IBWO-SVM model. By introducing refracted opposition-based learning and a cyclone foraging strategy, the convergence speed and global search capability of the IBWO are significantly enhanced. Firstly, KPCA is utilized to extract nonlinear principal component features from raw data. Subsequently, the SVM model is employed for fault classification. Experimental results demonstrate that the IBWO algorithm outperforms the Grey Wolf Optimizer (GWO), Whale Optimization Algorithm (WOA), Sparrow Search Algorithm (SSA), and the original Beluga Whale Optimizer (BWO). The KPCA-IBWO-SVM model achieves an average diagnostic accuracy of 95.86% , representing a 6.54% improvement over the KPCA-BWO-SVM model, thereby fully validating the effectiveness and application value of the proposed method. |
| Keywords: Improved Beluga Whale Optimizer Algorithm (IBWO); Support Vector Machine (SVM); fault diagnosis; Kernel Principal Component Analysis (KPCA); rolling bearing |
