Multiple Objective Co-Optimization of Switched Reluctance Machine Design and Control

This dissertation includes a review of various motor types, a motivation for selecting the switched reluctance motor (SRM) as a focus of this work, a review of SRM design and control optimization methods in literature, a proposed co-optimization approach, and empirical evaluations to validate the models and proposed co-optimization methods.

The switched reluctance motor (SRM) was chosen as a focus of research based on its low cost, easy manufacturability, moderate performance and efficiency, and its potential for improvement through advanced design and control optimization. After a review of SRM design and control optimization methods in the literature, it was found that co-optimization of both SRM design and controls is not common, and key areas for improvement in methods for optimizing SRM design and control were identified. Among many things, this includes the need for computationally efficient transient models with the accuracy of FEA simulations and the need for co-optimization of both machine geometry and control methods throughout the entire operation range with multiple objectives such as torque ripple, efficiency, etc.

A modeling and optimization framework with multiple stages is proposed that includes robust transient simulators that use mappings from FEA in order to optimize SRM geometry, windings, and control conditions throughout the entire operation region with multiple objectives. These unique methods include the use of particle swarm optimization to determine current profiles for low to moderate speeds and other optimization methods to determine optimal control conditions throughout the entire operation range with consideration of various characteristics and boundary conditions such as voltage and current constraints. This multi-stage optimization process includes down-selections in two previous stages based on performance and operational characteristics at zero and maximum speed. Co-optimization of SRM design and control conditions is demonstrated as a final design is selected based on a fitness function evaluating various operational characteristics including torque ripple and efficiency throughout the torque-speed operation range. The final design was scaled, fabricated, and tested to demonstrate the viability of the proposed framework and co-optimization method.

Accuracy of the models was confirmed by comparing simulated and empirical results. Test results from operation at various torques and speeds demonstrates the effectiveness of the optimization approach throughout the entire operating range. Furthermore, test results confirm the feasibility of the proposed torque ripple minimization and efficiency maximization control schemes. A key benefit of the overall proposed approach is that a wide range of machine design parameters and control conditions can be swept, and based on the needs of an application, the designer can select the appropriate geometry, winding, and control approach based on various performance functions that consider torque ripple, efficiency, and other metrics.