A prediction method for separating and quantifying noise contributions from casings and other plate like components in complex machines

In the context of noise source identification in machines, sound intensity, pressure or velocity distributions on a plane in space are commonly provided by intensity or holography maps. These maps, however, are not capable of separating the sources that are located on different planes, e.g. machines' radiating casings versus internal noise sources. The method discussed in this paper has been developed to analyze such situations. It consists of multiple partially coherent inputs–single output modeling to predict the radiations generated by machine casings or other plate like components. A parallel analysis with intensity will then enable to separate the casing or plate contribution from those of the internal sources. The application of this technique must, however, be accompanied by tools that allow to check its validity. Several of such tools are suggested. The method applications and validity checks have been discussed for a grinder wheel guard and a panel saw's blade casing.

► MPCISO applied to continuous media, i.e. plates, modeled as a set of partially coherent sources. ► Plate radiations prediction in presence of other noise sources, using vibration measurements. ► Smaller computation errors, compared to iterative spectral conditioning approach, through matrix inversion technique. ► A simulation method devised to identify the matrix singularities. ► ‘Impact radiation' technique examines the implementation and computation processes.