Enhanced Model Reference Adaptive Control based on Data-driven Technique for Permanent Magnet Synchronous Motors with Uncertainties

Precise velocity control of permanent magnet synchronous motors (PMSMs) is challenging due to the uncertainties induced by the mathematical model, unmodeled dynamics, and various disturbances; consequently, the traditional model reference adaptive control (MRAC) for PMSMs becomes questionable. In order to improve the traditional MRAC and lessen the influence of various sources of uncertainty, this study proposes an enhanced MRAC approach based on datadriven technique. First, the approach established an equivalent differential expression linear model for the discrete-time nonlinear description of PMSM and designed a data-driven controller using a model-free adaptive technique. This design depended on the unique bounded parameter referred as pseudo-partial-derivative (PPD) that was a slowly time-varying parameter relating to the system action point or system dynamics. Second, the approach identified the parameter PPD using Popov criterion to ensure the asymptotic stability of the controlled system. Finally, the approach was simulated and applied on a 50 kW PMSM drive of Toyota Prius II hybrid electric vehicles to demonstrate the effects of different control parameters. Results show that the proposed approach does not suffer from the drawback of the modeling process and unmodeled dynamics and provides improved velocity-tracking precision against parameter variations and external disturbances. When the weight factor increases from 0.1 to 1 or the pseudo-orders from 1 to 5, the velocity presents high robustness. Moreover, the approach yields satisfactory disturbance rejection and fault tolerance compared with the traditional MRAC even if some input and output (I/O) data are missing. This study solves the problem of dependence on plant models for the traditional MRAC and achieves better suppression of uncertainties, demonstrating the effective applications of this approach for actual nonlinear motor systems with uncertainties.