Source separation in diesel engines with the cyclic wiener filter
Résumé
The assessment and ranking of noise sources in Diesel engines is of major concern in the car industry. This is all the more true as the engine emission standards are more and more stringent, so that it is necessary to conciliate conflicting design modifications in order to reduce noise radiation on the one hand and engine emission on the other hand. In this context, there is a need for a fast and reliable method for separating the main engine noise sources (combustion noise and mechanical noise) from in operation measurements. This paper introduces such a method. It is based on advanced signal processing tools which explicitly take into consideration the non-stationary structure of the engine noise signals so as to design a cyclic Wiener filter to achieve separation. This results in a set of separated waveforms from which noise levels and frequency spectra relating to combustion noise and mechanical noise can easily be computed. It also allows the computation of the structural loss between the combustion chamber and the measurement points, and therefore the assessment of the contribution of each cylinder to the overall radiated noise. The proposed method has the advantage of requiring few instrumentation and of implying a low-complexity experimental protocol as compared to other existing solutions. Its application to some actual examples is illustrated.