Detection of inappropriate working conditions for the timing belt in internal-combustion engines using vibration signals and data mining

Abnormal operating conditions for the timing belt can lead to cracks, fatigue, sudden rupture and damage to engines. In this study, an intelligent system was developed to detect and classify high-load operating conditions and high-temperature operating conditions for timing belts. To achieve this, vibration signals in normal operating conditions, high-load operating conditions and high-temperature operating conditions were collected. Time-domain signals were transformed to the frequency domain and the time–frequency domain using the fast Fourier transform method and the wavelet transform method respectively. In the data-mining stage, 25 statistical features were extracted from different signal domains. The improved distance evaluation method was adopted to select the best features and to reduce the input space for the classifier. Then, the signal features from the time domain, the frequency domain and the time-frequency domain were fed into an artificial neural network to evaluate the accuracy of this designed procedure for detecting inappropriate operating conditions for the timing belt. Based on all these features extracted from the signals in the time, frequency and time–frequency domains, the artificial neural network classifier detected and classified normal operating conditions, high-load operating conditions and high-temperature operating conditions with accuracies of 73.3%, 85% and 89.2% respectively. The classification accuracies using features selected by improved distance evaluation in the signals from the time, frequency and time–frequency domains were found to be 85%, 95.8% and 95% respectively. The results showed that the developed system was capable of detecting and classifying both the normal operating conditions and abnormal operating conditions for the timing belt. The results also suggested that a combination of signal processing and feature selection can significantly enhance the classification accuracy.