A Model Based Virtual Sensing Approach for the Predictive Maintenance of Elevator Installations

Abstract:

The work in this dissertation addresses several open questions of the model based virtual sensing in its application to the predictive maintenance of elevators. Firstly, an electromechanical model for an elevator installation which comprises both the mechanical and electrical subsystems is developed in order to represent the dynamic behavior of an elevator during the ride. Secondly, a reduced-scaled elevator test bench is designed for the emulation of different elevator configurations. For this design, dimensional analysis is applied ensuring the similarity between the scaled test bench and the full-size system. Then, an extended state-space recursive least square algorithm for virtual sensing is derived.

The proposed extended state-space recursive least square algorithm is compared with the classical extended Kalman filter algorithm. Finally, key performance indicators are proposed for elevator installation predictive maintenance. The advantages and limitations of the model based virtual sensing approach are evaluated based on experimental tests in a scaled elevator test bench and in an elevator installation.