Automatic control procedure of the sound quality of engine starters in production
* Presenting author
Harmonic component analysis, click detection, study of the influence of shocks or resonance are basic tasks in acoustics, for numerous applications. Decomposing a complex signal in several single-component signals is a process which can often make further analysis tasks easier. GENESIS proposed a set of algorithms that allows the decomposition of an acoustic signal into four components: stationary noise, tonal part, transient part, remainder. For the sake of optimization, GENESIS proposed a sequential use of these computations, and integrated it in a procedure designed for facilitating the tuning task for the end user. In the context of quality assessments in automotive starter production lines, VALEO needs a reliable tool to detect and quantify the presence of high-frequency whistling with a method as non-intrusive as possible. A solution was found using an automatic component separation algorithm: tonal components are detected and extracted, allowing to estimate their contributions to the overall level with respect to the background noise. No tachometric measurement is required. Thus, an acceptation criterion is obtained, based only on the acoustic signal without any assumption over which orders are prominent.This paper describes the chosen algorithms and the tuning method that lead to the final implementation on the production line.