GA-Based Computer-Aided Electromagnetic Design of Two-Phase SRM for Compressor Drives

Abstract

The design optimization of a reluctance motor drive system for a given specifications is a multi-dimensional optimization problem. Achieving the design objectives often requires tens or hundreds of dimensional and electrical parameters to be tuned. To achieve efficient design, many researchers have suggested computer aided, autonomous, electromagnetic design approaches employing evolutionary computing, e.g., Genetic Algorithms (GA).As computer technology has progressed, Finite Element Analysis (FEA) has enabled easy and accurate analysis of all non-linear magnetic properties in terms of flux density and torque in the motors. However, it is still difficult to apply FEA for the repeatedcalculation required to obtain fitness evaluation in the computeraided, autonomous, electromagnetic design approaches using evolutionary computing because of long CPU times.This paper presents an approach to GA-based autonomous electromagnetic design of 2-phase Switched Reluctance Motor (SRM) drive. The application of the proposed drive is for compressor drives in refrigerators at household. From standpoints of the manufacturing cost, we have focused on 2-phase switched reluctance motor (SRM) as an alternative to the existing brushless DC motors with rare-earth magnets. Fig. 1 shows the schematic diagram of proposed computer-automated design approach. The proposed approach employs three GA loops. One achieves the lamination design optimization as shown in Fig. 1(a). The others are for autonomous finetuning of the control parameters required for the maximum torque or the high efficiency control required under frequent operating conditions (see Fig. 1(b)). To achieve the design optimization within an acceptable CPU-time, the repeated-calculation required to obtain fitness evaluation in the proposed approach does not use FEA, but consists of geometric flux tube-based non-linear magnetic analysis and a dynamic simulator based on an analytical expression of the magnetizing curves obtained from the non-linear magnetic analysis. As a result, the proposed analysis can significantly reduce in the computational time by 1/500 compared to 2D-FEA while keeping reasonable analytical accuracy.A 2-phase 8/6 SRM shown in Fig. 2 is designed using the proposed approach, which fulfills required torque characteristics and acoustic noise level under the dimensional restrictions and the available power limitations for the compressor drive. According to the optimum design results, a prototype 2-phase 8/6 SRM was constructed. The maximum torque curves measured in the prototype is shown in Fig. 3. The prototype can almost fulfill the maximum torques desired in the target application. In addition, the noise level of prototype in the frequent operating point L 1 shown in Fig. 3 is 4.7 dB lower than the desired value. From these results, the effectiveness of the proposed design approach is verified.