Identification of optimized diagnostic features of multi-stage gearbox condition

The occurrence of many gearbox faults simultaneously and possibilities of their identification on the basis of vibration symptoms is not a topic of research work in the literature. The authors of this research are concentrated on identification of a single type of fault on the basis of a single symptom. This paper presents the research work on selection of diagnostic features during gearbox degradation when many faults occur simultaneously. The paper presents the problem of choosing the signal receiving points. In the result of collected signal at many points we obtain a multidimensional matrix from which a point is to be selected for which we get the best distribution of symptoms for the particular faults. In these types of problems we try to find out the answer to the following questions: Which symptoms give the better results for classification of different faults at simultaneous different faults occurrence? For which signal receiving point good/bad technical state classes separate-ability is the highest? Is it possible to reduce symptoms matrix?.