Internal combustion engine sound-based fault detection and diagnosis using adaptive line enhancers

Abstract

In an internal combustion engine, the impulsive sounds are very often radiated owing to the faults of the engine. Thus it is important for a noise, vibration, and harshness engineer to detect and analyse impulsive sound signals for both fault diagnoses. However, it is often difficult to detect and identify impulsive signals because of interfering signals such as those due to engine firing, harmonics of crankshaft speed, and broadband noise components. These interferences hinder the early detection of faults and improvement of sound quality. In order to overcome this difficulty, a two-stage adaptive line enhancer which is capable of enhancing impulsive signals embedded in background noise is developed. This method is used to pre-process signals prior to time-frequency analysis via a bilinear methods such as the Wigner-Ville distribution and the Choi-Williams distribution.

An improved method of detecting engine misfire by sound quality metrics of radiated sound

“…These fault occurrences generate transient non-stationary sound/ vibration signals whose frequency-domain and time-domain features vary with time. [21][22][23] A traditional time-domain analysis or a frequency-domain analysis only provides a statistical average of the signal features in time/frequency domain. Hence, they are not able to accurately capture transient changes in frequency distribution and temporal patterns of a sound signal, caused by a fault occurrence.…”

“…In general, such engine sounds have major power distribution between 0 and 2000 Hz and negligible power beyond 2000 Hz. [21][22][23] Engine misfiring mainly affects the harmonics, and this change in frequency distribution will be at low frequencies (0-2000 Hz). Frequency weighting function applied to compute sharpness has very low values for 0-2000 Hz, and higher values at higher frequencies (see equation 2).…”

“…1,6,12,14,17 Research shows that engine and exhaust sound spectra change with change in engine operating conditions. [21][22][23] Research also suggests that engine and exhaust sound quality metrics may provide useful information about an engine's operating conditions. [21][22][23] Therefore, we propose the use of sound-quality metrics to diagnose the occurrence of a fault such as a cylinder misfire (initially proposed in 2017 by Singh et al 24 ).…”

“…[21][22][23] Research also suggests that engine and exhaust sound quality metrics may provide useful information about an engine's operating conditions. [21][22][23] Therefore, we propose the use of sound-quality metrics to diagnose the occurrence of a fault such as a cylinder misfire (initially proposed in 2017 by Singh et al 24 ). There is no other research that has suggested or tested the sound-quality metrics of the radiated sound, such as the one emanating from the engine or the exhaust, for misfire diagnosis.…”