System Identification of Lightly Damped Systems Using Recursive Kautz Functions-Case Study With Coriolis Mass Flow Meter

R Rajaprasanna,G Balasubramanian,K Ghousiya Begum

doi:10.1109/access.2021.3077608

R Rajaprasanna, G Balasubramanian + Show 1 more

Open Access

https://doi.org/10.1109/access.2021.3077608

Copy DOI

Abstract

This manuscript deals with the system identification of lightly (under) damped processes to find a suitable model structure. To identify such processes, the Kautz model, which is a two-parameter representation of orthogonal basis functions (OBF) in state space format, is used. The process parameters are obtained through the subspace method by sequencing the states obtained from the Kautz model. The obtained states are transformed into convex optimization in which the eigen values poles) to be estimated are confined to lie within user-defined regions in the convex plane. To derive the optimal states, an efficient control scheme based on the state observer design is implemented. The viability of the proposed work has been verified on real-time Coriolis Mass Flow Meter (CMFM) under no flow condition and the corresponding responses are plotted using MATLAB.

Highlights

Modeling of dynamic systems based on the system identification method is highly an iterative procedure
These challenges are the serious limitations in Impulse Response (IR) methods, even though, they are highly recommended to approximate any stable linear system as it guarantees stability and neglects truncation errors [2]
Coriolis mass flow meter is a type of flow measuring instrument that measures the flow in terms of mass

Summary

INTRODUCTION

Modeling of dynamic systems based on the system identification method (black box modeling) is highly an iterative procedure. Miller et al [13] proposed a novel method to estimate the system parameters by transforming the entire state space model into convex optimization with constraints in terms of Linear Matrix Inequality (LMI). The real-time or benchmark output is compared with the Kautz function output and the discrepancies are minimized using Euclidean norm analysis with constraints in terms of eigenvalues of the system matrix A1 and A2. It is a kind of quadratic optimization with non-linear constraints. The stability is guaranteed through the eigenvalues of the matrix AC in terms of unity circle constraint i.e., the eigenvalues should not go beyond the limits [-1;1]

ALGORITHM FOR GETTING THE OPTIMAL STATES AND PARAMETERS

REAL-TIME EXAMPLE

NUMERICAL RESULTS

CONCLUSION