A Comparative Analysis of Supervised Machine Learning Algorithms for Fault Prediction in Automotive Suspension Systems

Abstract

Intense automotive vibrations, while common, can seriously compromise driver health. This research utilizes machine learning (ML) to predict potential failures in car suspension systems, targeting an enhancement in vehicle reliability and safety. While numerous studies have simulated suspension faults, the increasing data complexity from uncertain parameters necessitates more efficient algorithms for precise fault identification. This study, therefore, conducts a comparative analysis of several supervised machine learning algorithms to determine the most accurate method for this predictive task. The algorithms were evaluated using four distinct feature set preparations: original data, standard deviation data, principal component analysis data, and a combined set of mean standard deviation and principal component analysis. The findings reveal that the Artificial Neural Network (ANN) and Support Vector Classifiers (SVC) algorithms yield the highest prediction accuracy. Notably, this peak accuracy was achieved when utilizing the combined feature set (mean standard deviation and Principal Component Analysis (PCA)). These results offer a valuable contribution toward designing more robust car suspension systems and advancing future preventive maintenance strategies.