A novel industrial wireless sensor network for condition monitoring and fault diagnosis of electrical machines

Abstract

A novel industrial wireless sensor network (IWSN) for condition monitoring and fault diagnosis of electrical machines is presented, in which on-sensor fault diagnosis based on principal component analysis is explored to address the tension between the high system requirements of electrical machine monitoring and the resource constrained characteristics of IWSN sensor nodes. The prototype system is evaluated with a single phase induction motor monitoring system. Normal motor working conditions and two types of motor faults, ie. loose feet and mass imbalance, are monitored to validate the feasibility of the proposed system. The results show that using on-sensor fault diagnosis can reduce transmission data by 99.8%, decrease energy consumption, and prolong node lifetime from 106 to 153 h, an increase of 44%. The experimental results also indicate that the proposed approach has high fault diagnosis accuracy.