Application of Combined Wavelet Transformation and Neural Network in Electrical System Malfunction Detection of Engine

Gohari, Mohammad; Pak, Abbas; Kazemi, Masoud; Rafieian, Farzad

doi:10.22064/tava.2025.2057650.1264

Application of Combined Wavelet Transformation and Neural Network in Electrical System Malfunction Detection of Engine

Document Type : Research Article

Authors

Mohammad Gohari ¹

Abbas Pak ²

Masoud Kazemi ³

Farzad Rafieian ⁴

¹ Associate Professor, School of Mechanical Engineering, Arak University of Technology, Arak, IRAN.

² Abbas Pak, Faculty of Mechanical Engineering, Engineering School, Bu Ali Sina University, Iran

³ Faculty of Mechanical Engineering, Arak University of Technology, Iran

⁴ Department of Mechanical Engineering, Arak University of Technology, Arak, IRAN.

10.22064/tava.2025.2057650.1264

Abstract

Engine failure is a critical issue for drivers and often requires substantial experience to diagnose and resolve effectively. Attempting repairs based on guesswork or uncertain causes can lead to significant time loss and high costs. Recently, Artificial Intelligence (AI) models, particularly those based on Artificial Neural Networks (ANNs), have shown promising performance in fault diagnosis. This study focuses on detecting two common faults in internal combustion engines—cylinder misfire and complete cylinder failure—both typically caused by problems in the ignition system. A model referred to as WANN (Wavelet+ Artificial Neural Network) is proposed, which uses coefficients are derived from vibration signals by wavelet transformation. The WANN achieved over 90% classification accuracy in identifying ignition-related faults. To evaluate the model's generalizability, the WANN is also tested on a different engine, successfully classifying fault types with acceptable accuracy. Notably, the model accurately detected ignition faults in a vehicle-mounted engine, demonstrating its robustness and practical utility.

This capability allows mechanics and technicians to accurately pinpoint the fault type, leading to more efficient and cost-effective repairs. Therefore, the proposed method offers a reliable and intelligent solution for diagnosing ignition system faults in automotive applications.

Keywords

Ignition System Malfunction

Wavelet Transformation

Artificial Intelligence Model

Diagnosis

Internal Combustion Engine

Subjects