Abstract
The use of optimization algorithms to design motor drive components is increasingly common. To account for component interactions, complex system-level models with many input parameters and constraints are needed, along with advanced optimization techniques. This article explores the system-level optimization of a motor drive design, using advanced evolutionary multiobjective optimization (EMO) algorithms. Practical aspects of their application to a motor drive design optimization are discussed, considering various modelling, search space definition, performance space mapping, and constraints handling techniques. Further, for illustration purposes, a motor drive design optimization case study is performed, and visualization plots for the design variables and constrained performances are proposed to aid analysis of the optimization results. With the increasing availability and capability of modern computing, this article shows the significant advantages of optimization-based designs with EMO algorithms as compared to traditional design approaches, in terms of flexibility and engineering time.
Highlights
The electric drive is an essential system in either industrial or mobile applications, since it handles the electromechanical power conversion [1, 2]
Modelling techniques include the use of analytical equations, lumped parameter models, and numerical analysis, while optimization algorithms are further classified into two types: deterministic, where the algorithm searches for solutions systematically, and stochastic, where it explores the design space randomly [3]
With drastic improvements in computing performance, different combinations of modelling techniques and optimization algorithms have been presented for the design of motor drive components, i.e. the machine and converter
Summary
The electric drive is an essential system in either industrial or mobile applications, since it handles the electromechanical power conversion [1, 2]. With the exception of [10], the mentioned publications do not consider the impact of design variables pre-selection, constraint handling techniques and performance space mapping methods on the optimization outcome These aspects are key for an effective use of the optimization algorithms, especially for complex problems such as system-level motor drive optimization. DE was originally designed for single objective problems, and to extend DE for multiobjective optimization problems, GDE3 is proposed in [17] This algorithm uses a Pareto dominance concept for its selection, along with a crowding distance index to ensure a diverse set of solutions. With the use of total constraint violation in (4) instead of maximum particular violation, no penalty factors are needed, and infeasible solutions are always quantitatively compared and penalized in a way such that they provide a search direction towards the feasible region [7]. We present a case study looking at using the GDE3 algorithm to optimize a motor drive system design, illustrating the concepts and theories discussed so far
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
