Adaptive online correction and interpolation of quadrature encoder signals using radial basis functions

This paper considers the development of an adaptive online approach for the correction and interpolation of quadrature encoder signals, suitable for application to precision motion control systems. It is based on the use of a two-stage double-layered radial basis function (RBF) neural network. The first RBF stage is used to adaptively correct for the imperfections in the encoder signals such as mean, phase offsets, amplitude deviation and waveform distortion. The second RBF stage serves as the inferencing machine to adaptively map the quadrature encoder signals to higher order sinusoids, thus, enabling intermediate positions to be derived. Experimental and simulation results are provided to verify the effectiveness of the RBF approach.