Statistical characteristics of electric drives at end-of-line test benches
* Presenting author
Abstract:
Newly built electrical drives have to be tested for faultless assembly and components. Over the last decades structure-borne noise testing has emerged as a valid option to check for faulty bearings, eccentric shafts or broken gears. In the context of highly automated series tests, exact limits have to be defined in order to be able to detect statistical deviations. Conventional algorithms for forming statistical boundaries are characterized by large blind spots or high reject rates . An effective statistical limit can be created by normalizing the distribution functions via a Box-Cox transformation. We show the characteristics of the distribution function of several hundred electric drives. Both normal and abnormal excitation mechanisms can be easily identified in the resulting Lambdagram. This opens up the possibility for the responsible engineerto expand excitation mechanisms of individually known theory with all other normal excitations present in his system.