The noise radiated by an ICE engine results from a mixture of various complex sources such as combustion, injection, piston slap, turbocharger, etc. Some of these have been categorized as combustion related noise and others as mechanical noise. Of great concern is the assessment of combustion noise which, under some operating conditions, is likely to predominate over the other sources of noise. The residual noise, produced by various other sources, is commonly referred to as mechanical noise. Being able to extract combustion and mechanical noise is of prime interest in the development phase of the engine and also for diagnostic purposes.
This paper presents the application of combustion mechanical noise separation techniques on a V8 engine. Three techniques, namely the multi regression analysis, the classical Wiener filter and the cyclostationary (synchronous) Wiener filter, have been investigated. The techniques have been applied to microphone recordings measured at one meter distance from an engine running on a test bench. Reference in-cylinder pressure sensors have been instrumented and synchronously acquired together with the microphones. Strengths and weaknesses of the techniques are assessed and presented in the paper.
Finally a source separation technique is applied as a pre-processing step to the sound recordings measured on a microphone array. Sound source localization then allows to localize and quantify the combustion and mechanical sources of noise on the engine. This is illustrated with an application example.