Torque Ripple Minimization of Switched Reluctance Drive Using a Neuro-fuzzy Control Technique

Simple power electronic drive circuit and fault tolerance of converter are specific advantages of SRM drives, but excessive torque ripple has limited its application. This paper presents a novel method of controlling the motor currents to minimize the torque ripple, using a neuro-fuzzy compensator. In the proposed control concept, a compensating signal is added to the output of a classical PI controller, in a current-regulated speed control loop. The compensating signal is learned prior to normal operation, in a self-commissioning run, but the neuro-fuzzy methodology is also suitable for online self-learning implementation, for continuous improvement of the compensating signal. I INTRODUCTION Many authors have proposed the dynamic control of a SR drive using fuzzy logic and neural networks [1-4]. This type of control is today well established in the area of motion control and particularly in drive systems. Artificial intelligence-based fuzzy, neural and fuzzy-neural controllers have a number of advantages over conventional controllers [5], and even helping to incorporate some "intelligence" into them [13-17]. The most remarkable advantages for SR Drives are: no requirement of an accurate model; possibility of design based exclusively on linguistic information derived from experts or from the use of clustering techniques and capacity of incorporation of new data and information as they become available by learning mechanisms. Fuzzy logic control of a SR drive has been implemented with success in [2], and has shown to be effective for the speed control in applications where some degree of torque ripple is tolerated, as is the case in many industrial applications [18]. Nevertheless, in servo control applications or when smooth control is required at low speeds, the elimination of the torque ripple becomes the main issue for an acceptable control strategy. In this case, even using a fuzzy PIlike control as the one described in [2] is not satisfactory, because the controller's output signal, which is used as a reference signal for the current control in the power converter, gives rise to sustained torque pulsations in steady-state. Furthermore, this torque ripple changes with the speed of the SR motor and with the load applied to it. II TORQUE PULSATION With a PI-like control alone, it is not possible to obtain a ripple-free output speed at any speed range, because it would also require a ripple-free output torque, for this purpose. If it is supposed that the output speed is constant and equal to the reference speed in steady-state, then the PI controller's output signal (i.e. the reference current) would be constant. However, a constant current reference would produce an oscillating torque (Fig. 1), rendering the ripple-free speed control unfeasible. The simulation results shown in Fig. 1 correspond to the current-regulated, full-load operation of a 750W SR motor, at rated speed (1800rpm). 0 2 4 6 8 10 12 14 16 0 1 2 3 4 5 6 t / ms T o rq ue / N m Figure 1 -Torque ripple produced by constant current reference signal (simulation). At high speeds, the torque pulsations would occur at higher frequencies, thus causing less speed ripple, due to the natural filtering provided by the mechanical load inertia. Furthermore, SR drives are usually operated in single-pulse mode at high speeds, without current control. In this case, the most effective way of reducing vibrations caused by torque pulsation is by way of turn-off angle control. At lower speeds, it is more convenient to compensate for the torque pulsations through phase current waveshaping. In this case, the current reference signal should vary as a function of position, speed and load torque, in order to produce the desired compensation. In fact, the optimum compensating signal is a highly non-linear function of position, speed and load. Several works [7-12] have been published, which use many different strategies to produce a compensating signal. Some authors [8,10] use the inverse of the static torque-current-position relationship, which are tabulated previously and stored in memory. However, this method is quite laborious and sensitive to parameter variations. In this work, a novel compensation method is proposed, which is based upon a self-tuning neurofuzzy compensator. The proposed compensation scheme is described in the next section. III PROPOSED METHOD Figure 2 presents a simplified block diagram of the SR-drive speed control system, showing the proposed neuro-fuzzy compensating scheme. The basic idea of the proposed method is illustrated in Fig. 3. The output signal produced by the compensator, comp I , is added to the PI controller's output signal, ref I , which should be ideally constant in steady-state but producing significant ripple, as shown in Fig. 3(a). The resulting signal after the addition is used as a compensated reference signal for the currentcontrolled SR drive converter, as shown in Fig. 3(b). The compensating signal should then be adjusted in order to produce a ripple-free output torque. PI Control ler + + Neuro-Fuzzy B lock Conver ter + Motor Ic o m p