Background: Piston rings are one of the most vital and critical components of ICEs. The purpose of the piston rings is to provide effective sealing so that the heat from the piston can be transferred to the cylinder wall and oil from the crankcase should be kept away from the combustion chamber. The wear of piston rings directly affects the efficiency and performance of the engine. As a result, to diagnose piston rings faults, an effective condition monitoring and fault diagnosis methodology must be adopted.Purpose: In this study, an attempt has been carried out to diagnose piston ring scuffing faults using vibration signature analysis.Methods: An accelerometer sensor was used to collect time domain vibration signals from the healthy and all faulty engine operating conditions. Spectral feature analysis techniques such as the fast Fourier transform (FFT), power spectrum density (PSD), and empirical mode decomposition (EMD) were used to post-process the acquired signals. Further, a neural network technique known as multi-layer perceptron (MLP) was employed to develop the model for the classification of piston ring scuffing faults. In the end, the outcomes of signal processing and machine learning techniques were evaluated for both engine operating conditions.Results: The results showed clear shreds of fault diagnostic information using FFT, PSD, and EMD analyses whereas the MLP technique provided the best neural network architecture with the least mean squared error value.Conclusion: The results obtained using the signal processing techniques can be considered as a reference for the diagnosis of various engine component faults viz. gudgeon pin, gears, bearings, camshaft, etc. Further, the classification models created with these signals may be utilized in the automobile sector to forecast faults in the previously stated engine components.