Calibration Of Accelerometer Parameters Based On Bilinear Model

With the increasing demand for the use of MEMS accelerometer,in order to improve the accuracy of the accelerometer,it is necessary to use software algorithms to calibrate the parameter errors of the accelerometer.Finding a simple,efficient and large-scale application of parameter calibration methods has become an important research direction.As we all know,temperature is an important factor affecting the accuracy of MEMS accelerometer.When the temperature changes,the intrinsic parameters such as zero offset and scaling factor of the accelerometer will also change accordingly.Therefore,modeling the nonlinear relationship between sensor internal parameters and temperature,and using temperature data as an input variable is an important means to simplify the calibration process and reduce calibration costs.Therefore,this paper explores the parameter calibration of the accelerometer based on the bilinear Hammerstein model.The main work and innovations are as follows:Firstly,in order to describe the nonlinear problem of sensor parameters changing with temperature,this paper uses the bilinear Hammerstein model to model the accelerometer parameters,so that the accelerometer has the ability to deal with temperature change scenarios,which is more in line with actual application scenarios.Secondly,a bilinear Hammerstein model of EIV(error-in-variables)is constructed for the fact that both the input data and the output data contain noise in the actual parameter calibration process.And a parameter estimation algorithm based on total least square method is proposed.In addition,considering that there are often multiple sensors in the actual scene,a distributed bilinear Hammerstein model parameter estimation algorithm based on total least squares is further proposed.Experimental results show that the proposed algorithm is significantly better than the existing estimation algorithm based on least squares,and the performance of the distributed algorithm is far better than the estimation performance of a single node.Finally,this paper solves the problem that the traditional calibration method cannot calibrate the accelerometer with only XY two axes.In the improved calibration method,a method of extending the twodimensional data to a three-dimensional attitude vector is proposed.In order to simplify the installation and operation process of calibration,a rotation invariant is proposed as the objective function of parameter calibration.And combined with the bilinear Hammerstein parameter estimation algorithm based on total least squares,an improved accelerometer calibration method is proposed,which effectively solves the problem of only XY two-axis accelerometer calibration,and is compatible with the compound calibration in the temperature changing environment.