FOG De-Noising Algorithm Based on Augmented Nonlinear Differentiator and Singular Spectrum Analysis

Abstract

A novel algorithm based on singular spectrum analysis (SSA) and augmented nonlinear differentiator (AND) for extracting the useful signal from a noisy measurement of fiber optic gyroscope (FOG) is proposed in this paper. As a novel type of tracking differentiator, augmented nonlinear differentiator (AND) has the advantages of dynamical performance and noise-attenuation ability. However, there is a contradiction in AND, i.e., selecting a larger acceleration factor may cause faster convergence but bad random noise reduction, whereas selecting a smaller acceleration factor may lead to signal delay but effective random noise reduction. To overcome the contradiction of AND, multi-scale transformation is introduced. Firstly, the noisy signal is decomposed into components by SSA, and the correlation coefficients between each component and original signal are calculated, then the component with biggest correlation coefficient is reserved and other components are filtered by AND with designed selection criterion of acceleration factor, finally the de-noising result is obtained after reconstruction process. There are mainly two prominent advantages of the proposed SSA-AND algorithm: (i) Compared to traditional tracking differentiators, better de-noising ability can be achieved without signal delay; and (ii) compared to other widely used hybrid de-noising methods based on multi-scale transformation, a parameter determination method is given based on the correlation coefficient of each decomposed component, which improves the reliability of the proposed algorithm.been explored to cancel out the random noise to improve the precision of temperature drift modeling and compensation. Multi-scale transformation method is a widely used technique for FOG de-noising, such as wavelet-based techniques [4,5], mode decomposition-based techniques [6][7][8], and singular spectrum analysis (SSA) [9,10]. By using multi-scale transformation methods, the signal can be decomposed from time-domain into frequency-domain, and then the high frequency noise can be extracted and eliminated, which are successfully applied for FOG de-noising especially for static signals, but fails in highly dynamic conditions. Optimal estimation method, such as Kalman filter, is another widely used technique for FOG de-noising [11,12]. To solve the divergent problem of Kalman filter, adaptive Kalman filters have been proposed [13][14][15], such as innovation based adaptive estimation, multiple models based adaptive estimation, and residual based adaptive estimation. Experimental results in literatures show that the Kalman filter based technique is promising for FOG de-noising both in static and dynamic conditions. However, the adjusting Kalman filter parameters for real-time filtering application still need to be improved. Moreover, many other de-noising methods have been reported for FOG in the literature, such as time-frequency peak filtering [16], forward linear predication [17], and hybrid de-noising algorithms [18]. These algorithms are available for FOG de-noisi...

Delay-Free Tracking Differentiator Design Based on Variational Mode Decomposition: Application on MEMS Gyroscope Denoising

“…Hence, CC is used here as a determination criterion for each R of STD. The steps of VMD-STD algorithm are given below [21].…”

Temperature Drift Compensation of Fiber Optic Gyroscopes Based on an Improved Method

“…The Allan variance analysis results indicated that the Q, N, and B reduced from 12.915 to 2.429 × 10 −1 , 1.8 × 10 −2 to 5.061 × 10 −4 , and 3.01 × 10 −1 to 1 × 10 −2 based on the X axis; from 7.680 to 1.38 × 10 −1 , 1.2 × 10 −2 to 2.647 × 10 −4 , and 1.72 × 10 −1 to 6 × 10 −3 based on the Y axis; and from 7.093 to 1.25 × 10 −1 , 1 × 10 −2 to Micromachines 2023, 14, 1712 3 of 18 1.549 × 10 −4 , and 1.53 × 10 −1 to 7 × 10 −3 based on the X axis, respectively. Zhang et al [20] proposed a novel algorithm based on singular spectrum analysis (SSA) and augmented nonlinear differentiator (AND) to extract the useful signal from a noisy measurement of FOGs. The proposed SSA-AND algorithm has a better denoising ability compared with other advanced denoising algorithms, and the temperature drift of the FOG can be extracted effectively without signal delay.…”

A New Quick-Response Sliding Mode Tracking Differentiator With its Chattering-Free Discrete-Time Implementation

“…That is chattering of system output due to the numerical error caused by inappropriate discretization of the discontinuous signum function [28], [29]. In order to attenuate chattering, some methods such as boundary layer [30], [31], continuous function [10], [11], adaptive switching gain [32], and high order sliding mode [5], [12] are applied. However, these methods reduce or avoid chattering by loosing the inherent characteristics of sliding mode technique in a strict mathematical sense.…”