Application of state estimation to the monitoring of multiple components in non-linear electro-mechanical systems

Abstract This paper presents a framework for condition monitoring of large non-linear electro-mechanical systems. Faults are assessed in different subsystems by means of model-based virtual sensors obtained through joint state-input estimation algorithms. Two outputs of the estimation are used as indicators of the condition of the system: (i) the estimated virtual sensors and unknown inputs, and (ii) the estimation innovations. These time series are converted into high-level key performance indicators that could be easily transferred and analyzed. The presented strategy is tested in a scaled test bench of a vertical transportation system.