CALIBRATION PROCESS OF TWO – AXES MATGNETOMETER BY NEURAL NETWORK

Abstract

The article presents an option how to design neural networks to solve the errors of 2D vector fields sensors: magnetometers and accelerometers. The introduction deals with the errors of the sensor. The theory describes that gradient method was used for learning corrective constants. The next, modelling, presents the development of corrective constants under the ideal conditions. We assembled, the real experiment confirms the theory and the modelling. The utilization of neural networks was directly shown on the improved course measuring.