Intelligent fault detection of spark plugs using vibration signal analysis and artificial neural networks

The primary concern pertaining to spark plug utilisation lies in their ignition efficiency and longevity. To address this issue, an intelligent and practical approach must be devised to serve as an indicator for determining the optimal time to replace a spark plug. In light of this, the present study is dedicated to the problem of fault detection in automobile engines, employing a sophisticated artificial neural network. The research utilises Time-Domain Signal Analysis as a means for extracting insightful characteristics, including RMS, kurtosis, skewness, and mean, from the data of vibration. Through the construction of an ANN model, distinct operating conditions simulating both healthy and faulty spark plugs are discerned. The vibration data acquisition is carried out utilising an accelerometer (ADXL335) interfaced with an Arduino Mega, which serves as the data acquisition device. The results demonstrate that the designed system exhibits an exceptional 100% overall accuracy in identifying faulty spark plug conditions. Copyright © 2024 Inderscience Enterprises Ltd.