A computationally intelligent maximum torque per ampere control strategy for switched reluctance machines

Abstract

While currently occupying only a niche role in industrial applications, the switched reluctance machines (SRM) unique operational characteristics could prove useful for additional engineering sectors given that inherent drawbacks are addressed. Phase winding isolation of SRMs provides greater fault tolerance when compared to the industrial standard, pulse width modulation driven induction machines. Furthermore, they may remain in a locked rotor position safely without concern of faulting and have higher speeds than many other electrical machines, i.e. contributing to greater overall robustness. When compared to other electrical machines, the SRM has higher currents requirements, creates greater acoustic noise and torque ripple, and requires more advanced controls for effective operation. Such drawbacks alienate the SRMs commercial and industrial popularity, ultimately limiting its full potential from being exploited. Since SRM torque production is typically non-linear, various techniques have been developed in order to maximize the torque output per unit current excitation, i.e. maximum torque per ampere (MTA). The “conventional” strategy, while simplistic, assumes a constant excitation over a symmetric period of the machine. This increases copper and iron losses while not effectively mitigating the current requirements or inherent torque ripple. By using particle swarm optimization (PSO), a stochastic search technique based on evolutionary algorithms, phase current MTA profiles may be obtained that optimize such conditions. This work presents a novel MTA SRM control strategy based on the PSO technique that obtains the optimum phase current profiles of a 4-phase, 8/6 pole SRM such that copper losses and torque ripple are minimized while achieving the desired torque at specific rotor positions.