Adaptive Neuro-fuzzy Controller for Multi-layered Switched Reluctance Motor

Abstract

ABSTRACT There has been big interest in switched reluctance motor (SRM) due to its simplicity and reasonable cost, however excessive torque ripple is one of the major disadvantages of switched reluctance motor. This paper attempts to reduce torque ripples of Switched Reluctance Motor through building multi-layered motor controlled by a hybrid intelligent system known as Adaptive Neuro-fuzzy Inference System ANFIS. Simulation of the proposed motor is conducted using Matlab Simulink environment 2011 and comparison results with single layer switched reluctance motor for both PI and ANFIS controllers show improvement in behavior of MSRM controlled by ANFIS through reduction in speed settling time as well as torque ripples. General Terms Electrical Machines; Adaptive Control. Keywords Rodrigues et Multi-layer switched reluctance motor; SRM; Torque ripples; ANFIS. 1. INTRODUCTION SRM has received attention due to its inherent simplicity, ruggedness, and low cost. These features make it a good candidate for various general purpose adjustable speed applications [1]. The most unique feature of SRM is its double saliency structure which means both stator and rotor has salient poles. The stator winding has concentrated coils, and shorter end turns than other types of motors, leading to a greater manufacturing economy. Moreover the salient rotor has no conductors or magnets, as shown in fig. 1. The rotor turns to get to a position of minimum reluctance by aligning itself with the stator magnetic field when the stator windings are excited, it is thus considered as the simplest of all electric machine rotors[1],[2]. Simplicity of the construction of SRM makes it inexpensive, the windings are electrically separate from each other which make the machine reliable, and cooling is simpler because the major sources of heat are on the stator. In addition, high speed capability of SRM and high torque to inertia ratio makes it a superior choice in different applications. However, Excessive torque ripple, especially at low speeds prevented SRM from widespread use [3]. Actually there are two approaches to reduce torque ripples: one of them is to improve the magnetic design of the motor, by changing the stator and rotor pole structures [4]. El-Kharashi [5] used the hollow cylindrical rotor to reduce torque ripples. There was no shaft, and the cylindrical rotor was designed to grade the air gap, and consequently the reluctance; which reduced ripples in the torque characteristic. Shang and Shing [6] changed the stator shape to c- shape one, the c-shape created more space for winding and hence became more straightforward, it had more number of turns per phase than traditional SRM; as a result the motor gave higher torque for the same current than in traditional SRM and consequently, minimize the cost of the motor. Daldaban and Ustkoyuncu [7] introduced multi-layer SRM. Their design allowed more space for stator winding and higher starting torque than classical SRM of the same size. The noise was reduced as well as torque ripples. The second approach is to use sophisticated electronic control. The electronic approach is based on selecting an optimum combination of the operating parameters, such as supply voltage, turn on and turn off angles and level of current [4]. al. [8] implemented fuzzy logic controller for reducing torque ripples through turn off angle. Zhang et al. [9] also used fuzzy controller to compute the current compensating signal that added to the main current value in order to reduce torque ripple. While Shang et al. [10] introduced flux-linkage controller using sliding mode technique (SM) with integral compensation (I). The integration of SM with I controller result in reduction of torque ripples when compared to each method alone.