Iterative Self Organized Data Algorithm for Fault Classification of Mechanical System

The challenging issue for mechanical industry is to develop fast & reliable fault diagnosis systems before total breakdown of machine. Fault diagnosis & classification of faults of mechanical systems is a difficult task. It improves productivity & reduces cost of production. This paper presents an approach for classification of commonly observed faults in gears of mechanical system. These faults include weared gear, gear with one tooth broken & gear with crack on one tooth. The Power Spectral Density (PSD) of the vibration signals of faulty gears is used to construct feature vectors. Principle component analysis (PCA) is used to reduce the dimensions of feature vector. The Routine checkup of the machine generates Known fault vectors. The ISODATA (Iterative Self Organizing Data Analysis Technique) [1] classifies fault vectors along with newly collected fault vector. If the fault is different from the previously identified fault a new fault cluster is created else new fault belongs to one of previously identified fault clusters.