Robust measurement of angular position using resolver sensor and ADALINE neural networks

This paper presents a new procedure to get the angular position of a motor shaft using a resolver sensor and artificial neural networks. The mentioned sensor is modeled as a linear system between the excitation and its output signals. Two real-time ADALINE neural networks estimate the regression coefficients of this model, and the angular position is obtained applying inverse trigonometric function. Simulation and experimental results proves that the proposed system has good accuracy and robustness against noise, using a simple mathematical structure, and it can be used in closed-loop control of electrical machines.