Research on sensor fault diagnosis of diesel generator sets based on LMD-SVM fusion algorithm

• Published in: Proceedings of SPIE, the international society for optical engineering Vol. 13664; pp. 136640Q - 136640Q-6
• Main Author: Bai, Lingling
• Format: Conference Proceeding
• Language: English
• Published: SPIE 16.07.2025
• ISBN: 1510692657; 9781510692657
• ISSN: 0277-786X
• DOI: 10.1117/12.3070647

As sensors in diesel generator units are prone to environmental interference, aging, and electrical noise, which lead to data distortion and compromise the safe operation of the units. Therefore, an efficient and accurate fault diagnosis method is urgently needed. In this paper, a novel intelligent diagnostic method is proposed, combining Local Mean Decomposition (LMD) with Support Vector Machine (SVM). Firstly, LMD is employed to extract key features from non-stationary signals, while feature vectors are constructed using Pearson correlation analysis and offset vector analysis. Then, an improved Particle Swarm Optimization (MPSO) algorithm is utilized to optimize the SVM classifier, enhancing classification accuracy and robustness. The results show that the overall classification accuracy of this method in 10 fault modes is 98.2%, which is significantly better than the traditional method.