希尔伯特-黄变换
支持向量机
随机性
粒子群优化
工程类
算法
模式识别(心理学)
分类器(UML)
人工智能
群体行为
计算机科学
数学
白噪声
电信
统计
出处
期刊:Measurement
[Elsevier]
日期:2023-04-25
卷期号:216: 112923-112923
被引量:18
标识
DOI:10.1016/j.measurement.2023.112923
摘要
To extract representative fault features from the wheel-rail vibration signals caused by metro wheel faults, and to make the result of classification achieve the best balance between accuracy and time consumed, this study proposes a wheel fault identification method, which combines Ensemble Empirical Mode Decomposition (EEMD), Multi-scale Permutation Entropy (MPE) and Quantum-behaved Particle Swarm Optimization Support Vector Machine (QPSO-SVM). EEMD is used to make time–frequency decomposition of the signal adaptively according to the local characteristics of the signal itself. MPE is used to measure the complexity and randomness of the reconstructed signal at different scales. The QPSO-SVM classifier is proposed for pattern recognition. The results of simulation and engineering experiments show that the QPSO-SVM saves up to 80.8% of the classification time compared with the existing classifiers, and the proposed method can be used efficiently and maintain high accuracy in several working conditions common in metro operation.
科研通智能强力驱动
Strongly Powered by AbleSci AI