Abstract
A magnetic field-based sensing system utilizing statistically optimized concurrent multi-sensor outputs for non-contact precise field-position association is presented. The rationale and principle of capitalizing on simultaneous spatial field measurements to induce unique correspondence between field and position to achieve accurate translational motion over large travel distances for feedback control is illustrated using a single-source-multi-sensor configuration. Principal component analysis (PCA) is used as a pseudo filter to optimally reduce the dimension of the multi-sensor output space for field-position mapping with artificial neural networks (ANNs). The effects of PCA on the sensing accuracy and closed-loop tracking performance are experimentally investigated using a voice-coil motor and a 9 sensor network with an optical encoder as a comparison.