Identification of unknown parameter value for precise flow control of Coupled Tank using Robust Unscented Kalman filter

Abstract

In this paper, we consider the problems of state estimation and parameter estimation. The goal is to consider Robust Unscented Kalman filter, and demonstrate their successful application on a Coupled Tank system. Traditional unscented kalman filter have a limitation to estimate the state and parameter of time-varying parameter system due to making use of fixed measurement covariance without updating measurement error between measured data and estimated data. Proposed method is Robust Unscented Kalman filter to perform the estimation of the changing parameter value. A structure of the Coupled Tank System consists of connected two tank with basin. The other goal is to make use of the considered filtering method to compare between the other methods. Extensive experiments by numerical simulations and experimentation using real hardware are performed. The study of the experimental results shows a proposed method concern various aspects, such as estimation accuracy, convergence speed, and the accuracy of estimating fixed parameter values. Overall, the proposed Unscented Kalman filter turned out the best of the other considered methods.