Optimization of Winding and Rotor Harmonics in Variable Reluctance Resolvers Using Bayesian Techniques

Rotational resolvers are among the most important position sensors and are generally classified into two main categories: wound rotor and variable reluctance types. Although wound rotor resolvers offer higher accuracy, variable reluctance resolvers are widely favored due to their high reliability and simple manufacturing process. In this paper, a variable air-gap resolver is studied, and both the winding arrangement and rotor contour are optimized to enhance its accuracy to the level of, or even beyond, that of a wound rotor resolver. In this regard, various winding arrangements are analyzed including CECS (Concentrated Excitation and Concentrated Signal winding), CETS (Concentrated Excitation and Toroidal Signal), TECS (Toroidal Excitation and Concentrated Signal) and TETS (Toroidal Excitation and Toroidal Signal), and the one that yields the highest accuracy is selected. The impact of the number of rotor pole pairs on accuracy is also examined. Finally, harmonic components are introduced into the rotor contour to mitigate the harmonic content present in the signal voltages. The amplitude of the injected harmonics is optimized using Bayesian Optimization and Gaussian Process Regression (GPR), demonstrating that the resolver’s accuracy can reach -or even surpass- that of a wound rotor resolver. The optimization results are subsequently validated through prototyping.