Abstract
The performance of advanced drive systems in most of the sophisticated motion control industries is compromised by inherent sensor noise, which, in extreme cases, may induce undesirable instability. This research investigates the application of a Kalman filter-based model predictive controller to attenuate noise generated by the mechanical speed sensor in a 5-phase field oriented controlled permanent magnet synchronous machine drive system. Comparative analysis with a proportional-integral-based controller reveals that the proposed method can potentially reduce both rotor speed and electromagnetic torque noise by approximately 75% respectively under steady-state conditions. And, stator current is reduced by 66% with proposed controller. Initial evaluations are conducted in an offline MATLAB/SIMULINK environment, a crucial procedural step. Despite yielding accurate results, the substantial computational burden due to sequential processing, estimated at 1900 (approx.), poses challenges for its practical implementation. To address this limitation, the offline models are converted to online models with a Field Programmable Gate Array based rewritable OPAL-RT [4510] real time simulator with its enhanced parallel processing capability. This cost-effective methodology not only substantiates the research findings by eliminating system-specific and exorbitant experimental configurations but also augments the practicality of real world applications by diminishing the computational burden to the requisite value of 1.
Keywords: 5-phase PMSM, Drive System, Field Oriented Control, MPC, Real-time Simulation