Full ceramic bearing fault diagnosis using LAMSTAR neural network

Abstract

In this paper, an integrated full ceramic bearing fault diagnostic system developed with acoustic emission (AE) sensors and a large memory storage and retrieval (LAMSTAR) artificial neural network (ANN) is presented. LAMSTAR is a newly developed and US patented neural network algorithm. The performance of the diagnostic system is compared with those implemented with other types of fault classification algorithms using laboratory seeded fault test data. The presented diagnostic system with LAMSTAR network achieved over 93% individual fault detection accuracies along with over 96% overall accuracy.