Design optimisation of electrical motors by an adaptive model based optimisation technique

A new adaptive model based optimisation (AMBO) technique, particularly suitable for the optimisation of electrical motor design, has been proposed in this paper. It is based on an appropriate modelling of the objective function, which is adapted and improved during the optimisation process. Any effective optimisation algorithm can then be applied to the approximated objective function. The novel technique is deeply explained and simple examples of application are at first described in order to highlight the peculiar differences between AMBO and the already known procedures. Then it has been applied to the design of an electrical motor, whose analysis is at first performed by an analytical way and then by a finite element approach. Good results have been obtained in both the cases pointing out that the AMBO technique is an effective tool for the motor design optimisation. In particular, it allows the computation time to be greatly reduced since it requires a limited number of calls to the motor analysis procedure. Such a feature makes practicable the use of accurate finite element methods of analysis that otherwise are dramatically time costly. Moreover, the harmful effects of the local optima, can be widely mitigated, making easier the search of the global optimum of the design