Design and Calibration of a Novel Angular Position Sensor Based on BP Neural Network

Abstract

Based on the law of electromagnetic induction, a novel contactless absolute angular position micro-sensor has been designed in this paper. It is mainly composed of three parts: a magnet having two portions shaped as circle segments with different center points, a Hall effect sensor and a signal processing circuit. The linearity of the sensing system is improved by particularly designing spatial magnetic field distribution of asymmetric magnet, and Hall output signal is converted by the second-order low-pass filter circuit into linear analog voltage. Experiment has been performed on a prototype of the angular position sensor and the results demonstrate the proposed scheme is feasible. Theoretic precision and testing error analysis of the sensor has been pursued based on BP neural network at the end of the paper. The error does not exceed ± 0.015°. The prediction result can fully reproduce the output performance of hall sensing system with high accuracy of 0.02%, achieving high-precision angle measurement.