Extended-Kalman-filter-based dynamic mode decomposition for simultaneous system identification and denoising

TL;DR

This paper introduces an extended Kalman filter-based dynamic mode decomposition method that enables real-time system identification and denoising, especially effective for many-degree-of-freedom problems with system noise.

Contribution

The paper presents EKFDMD, a novel online DMD approach combining extended Kalman filtering and truncated POD, improving system identification and denoising for complex noisy systems.

Findings

EKFDMD estimates eigenvalues accurately from noisy data.

EKFDMD effectively denoises data online.

EKFDMD with trPOD handles many-DoF problems successfully.

Abstract

A new dynamic mode decomposition (DMD) method is introduced for simultaneous online system identification and denoising in conjunction with the adoption of an extended Kalman filter algorithm\color{black}. The present paper explains the extended-Kalman-filter-based DMD (EKFDMD) algorithm and illustrates that EKFDMD requires significant numerical resources for many-degree-of-freedom (many-DoF) problems and that the combination with truncated proper orthogonal decomposition (trPOD) helps us to apply the EKFDMD algorithm to many-DoF problems. The numerical experiments of the present study illustrate that EKFDMD can estimate eigenvalues from a noisy dataset with a few DoFs better than or as well as the existing algorithms, whereas EKFDMD can also denoise the original dataset online. In particular, EKFDMD performs better than existing algorithms for the case in which system noise is present. The EKFDMD with trPOD can be successfully applied to many-DoF problems, including a fluid-problem example, and the results reveal the superior performance of system identification and denoising. Note that these superior results are obtained despite being an online procedure.