On acoustic measurement-based condition monitoring of internal combustion engines

Abstract

Internal combustion engine acoustic signatures are rich in information about the engine's operating parameters and physical condition. Unfortunately, these signatures are very complex and highly corrupted by background noise, therefore the extraction of diagnostic signatures is difficult and not straightforward. This article focuses on ways to extract quick information about the internal combustion engine's operating conditions by means of their induced acoustic signals in a normal, acoustically untreated laboratory environment, without any sound measurement precautions. The engine acoustic sources are briefly outlined and their fundamental characteristics are studied in time domain, frequency domain, and other statistical parameters analyses. Acoustic signals for a range of loads and speeds are considered in order to provide a baseline for normal engine characteristics. An analysis of a number of measured parameters from the engine test rig is then given for the purposes of detecting, diagnosing and assessing the severity of some seeded faults. Namely, these faults are decrease in compression ratio, change in injection pressure, and variation of exhaust and intake valve's clearances. Exhaust valve clearance-related faults were also detected and diagnosed by monitoring the spikeness level using the r.m.s. values of the acoustic signals kurtosis of each of the engine cylinders. Conclusions and concluded marks are also given.