Tuning of a type-III software-based resolver-to-digital converter through genetic algorithm

Resolver is a sensor used in applications that require a reliable angle measurement. However, getting the angular position from the resolver outputs is a difficult task. Observers called resolver-to-digital converters (RDCs) are used to get the angular position from resolver. In many researches, RDCs are estimation algorithms (software-based RDCs). The transient response and robustness against noise of the angle estimation depends on the parameters of the RDC. This paper proposes the use of a genetic algorithm (GA) to set the parameters of a type-III software-based RDC. GA is a heuristic algorithm that search a solution that minimize a fitness function. In this paper, the solution is composed by the gains of the angular tracking observer (ATO) that composes the RDC. On the other hand, the cost function is a linear combination of the error peak, the settling time and the effect of noise in the angle estimation. Simulations shows that the angle estimation using the proposed approach has accuracy, good transient response and robustness against noise. Besides, GA allows customizing the ATO in order for the RDC to give an estimation where the noise rejection or the settling time is priority.