Title: Adaptive cubature quadrature filter for nonlinear state estimation

Authors: Aritro Dey

Addresses: Department of Electrical Engineering, National Institute of Technology, Durgapur, Durgapur, 713209,West Bengal, India

Abstract: A new filtering algorithm has been proposed for nonlinear state estimation where the measurement vector is a nonlinear function of system states and measurement noise. The proposed adaptive cubature quadrature filter demonstrably presents improved estimation performance in the situation where the measurement noise covariance remains unknown to the designer. The filter has been designed based on Bayesian filtering framework with cubature quadrature rule for approximation of Gaussian integral and also incorporates adaptation algorithm designed for auto-tuning of unknown measurement noise covariance. The adaptation algorithm, theoretically developed following maximum likelihood estimation (MLE) for non-additive noise, is numerically stable as it secures the positive definiteness of adapted measurement noise covariance. The superiority of the proposed filter has been demonstrated in simulation over its non-adaptive counterpart and the competing algorithms of adaptive nonlinear filters with the help of some non-trivial case studies. Additionally, suitability of the proposed algorithm is also validated for non-stationary measurement noise.

Keywords: adaptation; cubature quadrature filter; MLE; maximum likelihood estimation; measurement noise covariance; non-additive noise; nonlinear filtering.

DOI: 10.1504/IJMIC.2020.116912

International Journal of Modelling, Identification and Control, 2020 Vol.36 No.3, pp.211 - 221

Received: 09 Apr 2020
Accepted: 25 Sep 2020

Published online: 01 Aug 2021 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article