Automated Multi-Objective Design Optimization of PM AC Machines Using Computationally Efficient FEA and Differential Evolution

Exploiting cloud computing in the multi-physics design and optimisation of electromagnetic devices

“…Alternative optimisation methods could identify a Pareto front in a more robust and efficient manner, [4], [27], [29]. Adding compute resources using the simple algorithm presented here, Section III, accelerates the design development cycle and can allow a more thorough evaluation of the design space within a short time-frame, at a moderate cost and can potentially lead to improved designs.…”

“…Hence this method is computationally inefficient and tends to result in a low quality approximation to the Pareto front, where the Pareto points are bunched close together, [27]. A Pareto set can be identified more efficiently using alternative optimisation techniques such as Differential Evolution, [4], or Non-dominated Sorting Genetic Algorithm II (NSGA-II), [29] and can reduce both time and economic cost. Implementing such a change in optimisation approach would require the inductor model, Fig.…”

“…This significantly increases computation time and tends to prohibit the use of such models, particularly for iterative design optimisation. The problem is often addressed by employing computationally efficient modelling methods, [4], or by compromising the detail and accuracy of the models, [2], [5]- [7], in favour of short solution times. Thereby, enabling a design optimisation to be performed using commonplace desktop computing hardware, [4], [7].…”

“…The problem is often addressed by employing computationally efficient modelling methods, [4], or by compromising the detail and accuracy of the models, [2], [5]- [7], in favour of short solution times. Thereby, enabling a design optimisation to be performed using commonplace desktop computing hardware, [4], [7]. An alternative is to adopt a distributed computing approach, [8], where high-fidelity models are evaluated using many networked computers in parallel to reduce solution times.…”

Establishing the Relative Merits of Interior and Spoke-Type Permanent-Magnet Machines With Ferrite or <roman>NdFeB</roman> Through Systematic Design Optimization

“…[4][5]6,7,8,9,10,11 Previous studies have reported that DOE methods are more effective when the numbers of design variables and candidate designs are relatively small 5,6 and that DE algorithms are recommended for large scale numbers of variables and candidate designs. 7,10,11 In this paper, a combined DOE and DE optimization method was developed for the comparative study of six types of PM machines, namely, Parametric models for these generalized topologies, which allow geometrical morphing in between a spoke and a flat-bar IPM configuration, are introduced in the next section. The general mathematical formulation of the optimization problem and the general procedure, which combines DOE and DE, are presented in Section III.…”