An application of acoustic measurements to quality control of low power electrical motors

Abstract

In this article, an example of the application of acoustic measurements to condition assessment of electric machines is presented. Quality control of new components is considered in this case study. The first symptoms of unserviceability are disclose in the acoustic domain and then, in later exploitation, they develop into a form that is detectable during vibration measurements. The assessment is based on acoustical wave spectrum normalized with respect to rotational speed. An adequately prepared neural network of the Kohonen's type was used as assessing tool. Its performance was compared to results of the minimum distance method based on averaged spectrum patterns.