Fault diagnosis and prognosis by using Input-Output Hidden Markov Models applied to a diesel generator

In this paper, a method to implement a platform of failure diagnosis and prognosis, and health monitoring based on data using Input Output Hidden Markov Models (IOHMM) is proposed. Several sensors on a diesel generator system give information such as on-line operating conditions. The goal of this work is to use on-line collected data in order to determine degradation state of the diesel generator system. Classification of data thanks to IOHMM allows the monitoring of the component health state and operates fault diagnosis and prognosis.