Improved K-means Algorithm for Fault Diagnosis of Vehicle
Dongsheng Yang, Jiawei Xie, Zhenyu Yin · 2023
This paper aims to improve the accuracy of anomaly data identification in fault diagnosis of vehicle using an improved K-means clustering algorithm. To address the sensitivity of initial cluster centroids and the tendency to fall into local optima when identifying anomaly data with traditional K-means clustering algorithms, we propose an IDBO-Kmeans clustering algorithm for vehicle fault diagnosis. The algorithm is based on the original Dung Beetle Optimization algorithm and introduces three optimization strategies: chaotic mapping, adaptive boundary adjustment, and random walk, to improve the quality of initial cluster centroids. The optimized initial cluster centroid are then used with the K-means algorithm for further clustering. Experimental results show that IDBO-Kmeans improves the accuracy of anomaly data identification in vehicle fault diagnosis and addresses the limitations of traditional K-means clustering.