The Indirect Estimation Algorithm of MEMS IMU Stochastic Error Properties Based on Wavelet Variance

Abstract

As the development of micro-electromechanical systems (MEMS), the cheap and small dimension MEMS IMU has been widely used in the navigation system. However, MEMS IMU presents different stochastic errors, which are difficult to be analyzed and may degrade the accuracy of the navigation systems in a short period. Considering the disadvantages of the Allan variance method in stochastic error property analysis, the indirect estimation method is proposed in this paper. By Daubechies discrete wavelet transform for the backward differential stochastic error, the statistical characteristics of wavelet coefficients is thoroughly studied. The wavelet decomposition scale is determined by wavelet coefficient statistical characteristics. Then wavelet variance is selected as the auxiliary parameter of indirect estimation, and an optimal criterion of asymptotic consistency is derived. Finally, according to the relationship between stochastic error statistical characteristics and wavelet variance, the stochastic error property parameters with asymptotic consistency are obtained by the nonlinear Gauss-Newton method. Comparing with estimated results by Allan variance method, simulation results indicates that the indirect estimation method not only improves the accuracy of parameter estimation, but also effectively resolved the issue concerning accurate parameter estimation of a first-order Markov stochastic error model.