Chengbo Yang, W Y Liu, Bao Song, K. T. Chau
Temperature-dependent electrical parameters, i.e., stator resistance and rotor flux linkage, are crucial for the control performance optimization and thermal safety management of surface-mounted permanent magnet synchronous motor (SPMSM) drives. In this paper, an alternating dual-Adaline observer network (DAON)-based method is proposed to enable the signal-injection-free online identification of rotor flux linkage and stator resistance. It employs a cyclically operating structure consisting of two DAONs, which are designed for alternating activation and mutual updates, thereby circumventing the rank-deficient issue under insufficient excitation conditions and counteracting the inverter nonlinearity. Each DAON incorporates two Adaline-type sub-observers, both of which feature a tailored self-regulating learning rate designed via Lyapunov stability theory. Such a learning rate design not only guarantees the asymptotic convergence of the identification error but also provides an explicit selection range for the to-be-adjusted parameter, greatly enhancing engineering reliability and practicality. Simulations, as well as real-time experiments conducted on a 1.2-kW SPMSM drive system, demonstrate that the presented method is capable of delivering high-accuracy identification for the two temperature-dependent parameters, irrespective of the excitation level.