MACHINE LEARNING-BASED ACOUSTIC MONITORING FOR EARLY RECOGNITION OF ENGINE KNOCKING AND VEHICLE FAULT IDENTIFICATION

Abstract

The research paper investigates machine learning as a tool to detect engine knocking in real time to improve the process of recognizing fault in the vehicle in an early stage. The frequency modulation amplitude demodulation (FMAD) engine-sound-data features were extracted and several machine-learning algorithms were assessed through MATLAB. Coarse decision tree algorithm proved to be the most efficient with accuracy in classification standing at 66.01%.Additionally, we introduced a knock index to quantify noise levels during each engine cycle. This index, calculated from the integral of the absolute value of the first derivative of a band-pass-filtered vibration signal, provides a visual representation of knock strength. By comparing the knock index to a statistically defined threshold, we could distinguish between normal and knocking cycles. Our results demonstrated consistency between experimental observations and knock index characteristics. This approach shows promise for early detection of engine knocking, although further refinement of feature extraction methods and algorithm optimization is necessary for practical application. The study highlights the potential of integrating machine learning into real-time vehicle fault detection systems to improve their reliability and effectiveness.