Implementation and evaluation of online system identification of electromechanical systems using adaptive filters

Online system identification method provides useful insight into the electromechanical plant behavior during commissioning phase as well as health monitoring. In addition to controller tuning, the methods can be applied for identifying un-modelled mechanical resonances, sensor vibrations, bearing friction, etc. Accordingly, this paper focuses on the implementation and evaluation of real time frequency response of electromechanical systems using adaptive filters. Some of the challenges during implementation include selecting the structure of adaptive filters and understanding influence of convergence gain on accuracy of frequency response. In this work, online system identification is demonstrated on two setups (a) magnetic suspension system and (b) sensorless permanent magnet motor drive. The experimental result validate that the method elaborated is simple, easy to implement and accurate when compared with commercial off the shelf spectrum analyzer.