Rolling bearing fault diagnosis based on Deep Boltzmann machines

Abstract

Rolling bearing is one of the most commonly used components in rotating machinery. It is easy to be damaged which can cause mechanical fault. Thus, it is significance to study fault diagnosis technology on rolling bearing. This paper presents a Deep Boltzmann Machines (DBM) model to identify the fault condition of rolling bearing. A data set with seven fault patterns is collected to evaluate the performance of DBM for rolling bearing fault diagnosis, which is based on the health condition of a rotating mechanical system. The features of time domain, frequency domain and time-frequency domain are extracted as input parameters for the DBM model. The results showed that the accuracy presented by the DBM model is highly reliable and applicable in fault diagnosis of rolling bearing.