Detection of engine valve faults by vibration signals measured on the cylinder head

Abstract

This technical note proposes a simple technique for the detection of incipient engine valve faults by vibration signals measured on the cylinder head. The characteristics of the vibration signal are analysed, indicating that its time domain and frequency domain characteristics are both useful for engine diagnosis while the cycle-by-cycle variation seems a disadvantage. A simple diagnostic technique named partial sampling and feature averaging (PSFA) has been presented. Only a specific part of the vibration signal corresponding to a particular impact force within each operating cycle is sampled and analysed, and then the diagnostic features are extracted from each part but averaged among many cycles. Identification of abnormal valve clearance just requires a partial sampling of the vibration signal extracted during the period of valve closing, and detection of gas leakage from valves needs a partial sampling signal corresponding to the period of in-cylinder combustion. The experimental results show that the proposed technique is feasible, effective and simple in implementation.