Torque ripple minimization in non-sinusoidal synchronous reluctance motors based on artificial neural networks

Phuoc Hoa Truong (1) , Damien Flieller (2) , Ngac Ky Nguyen (3) , Jean Merckle (1) , Guy Sturtzer (4)

1 MIPS - Modélisation, Intelligence, Processus et Système
2 GREEN Laboratory - INSA de Strasbourg
3 L2EP - Laboratoire d’Électrotechnique et d’Électronique de Puissance - ULR 2697
4 LGeco - Laboratoire de Génie de la Conception

This paper proposes a new method based on artificial neural networks for reducing the torque ripple in a non-sinusoidal synchronous reluctance motor. The Lagrange optimization method is used to solve the problem of calculating optimal currents in the d–q frame. A neural control scheme is then proposed as an adaptive solution to derive the optimal stator currents giving a constant electromagnetic torque and minimizing the ohmic losses. Thanks to the online learning capacity of neural networks, the optimal currents can be obtained online in real time. With this neural control, each machine's parameter estimation errors and current controller errors can be compensated. Simulation and experimental results are presented which confirm the validity of the proposed method.

Mots clés: Non-sinusoidal synchronous reluctance motor; Torque ripple; Optimal currents; Lagrange optimization; Adaline; Artificial neural networks

Domaines: Autre

Phuoc Hoa Truong, Damien Flieller, Ngac Ky Nguyen, Jean Merckle, Guy Sturtzer. Torque ripple minimization in non-sinusoidal synchronous reluctance motors based on artificial neural networks. Electric Power Systems Research, 2016, 140, pp.37-45. ⟨10.1016/j.epsr.2016.06.045⟩. ⟨hal-01511312⟩