Inertial Measurement Units' Error Calibration Based on Neural Network

A multilayer feed forward neural network(NN)was designed to compensate the Inertial Measurement Units(IMU)' non-linear errors.The neural network(NN)is good at approximating the nonlinear function,and thus suitable for the modeling of the non-linear system.As main approaching tool,back propagation neural network was used to approximate the non-linear error function.This method overcomes the disadvantage of traditional modeling method.Its application in modeling a MEMS IMU's non-linear error shows that the identification error of the model is tolerable in application.The experimental results show that this method can reduce the measurement errors of IMU effectively.