MEMS Accelerometer Noises Analysis Based on Triple Estimation Fractional Order Algorithm.

Michal Macias,Wiktor Malesza,Dominik Sierociuk

doi:10.3390/s22020527

Michal Macias, Wiktor Malesza + Show 1 more

Open Access

https://doi.org/10.3390/s22020527

Copy DOI

Journal: Sensors (Basel, Switzerland)	Publication Date: Jan 11, 2022
Citations: 8	License type: CC BY 4.0

Affiliation: Warsaw University of Technology

Abstract

This paper is devoted to identifying parameters of fractional order noises with application to noises obtained from MEMS accelerometer. The analysis and parameters estimation will be based on the Triple Estimation algorithm, which can simultaneously estimate state, fractional order, and parameter estimates. The capability of the Triple Estimation algorithm to fractional noises estimation will be confirmed by the sets of numerical analyses for fractional constant and variable order systems with Gaussian noise input signal. For experimental data analysis, the MEMS sensor SparkFun MPU9250 Inertial Measurement Unit (IMU) was used with data obtained from the accelerometer in x, y and z-axes. The experimental results clearly show the existence of fractional noise in this MEMS’ noise, which can be essential information in the design of filtering algorithms, for example, in inertial navigation.

Highlights

Micromachined Electrical Mechanical Systems (MEMS) are mechanical and electromechanical devices made using microfabrication techniques
The paper presents the results of the Triple Estimation Algorithm application to estimate fractional order noises
Presented numerical experiments show the ability of the algorithm to determine the variable order for the fractional variable order (FVO) noise

Summary

Introduction

Micromachined Electrical Mechanical Systems (MEMS) are mechanical and electromechanical devices made using microfabrication techniques. A crucial area in which MEMS sensors are used is inertial navigation systems (INS) [6,7] based on double integration of body acceleration processes based on accelerometers and gyroscopes measurements. Due to the double integration action, high accuracy and precision of acceleration measurement are essential because noises (especially biases) are double-integrated and rapidly increase navigation errors. That is why modelling noises, biases, and general dynamics of MEMS sensors are essential. An article [8] uses, for example, an advanced type of recurrent neural network to model some parts of nonmodeled MEMS gyroscope dynamics and apply this network into fractional order sliding mode control.

Methods

Results

Conclusion