Soft sensor based estimation of process parameters and states with Hybridized Grey Wolf Optimizer

Metaheuristics based global optimization technique is one among the state of art in soft sensing applications. The metaheuristic based soft sensor is capable of coping with the stochasticity of the process and as well as the corresponding nonlinearities in the dynamics. In this work, a multi-objective based parameter estimation and soft-sensing of unknown states are carried out for a modified version of Quadruple Tank Process (QTP). The effect of the outlet valve over the dynamics of the process is briefly investigated to emphasize the implication of these parameters in the process. The unknown parameters and unobserved states of the QTP are estimated through a Hybridized Grey Wolf Optimization (HGWO). The HGWO is a metaheuristic based optimization, hybridized with static Kalman Bucy (KB) algorithm. The hybridization is carried out to improve the convergence rate of the existing algorithm in terms of ideal computational cycle and proximity towards the global solution. Performance of the proposed HGWO is compared along with the conventional GWO algorithm. The estimation is executed for QTP operating in Non-Minimum Phase (NMP) mode and the simulated results proves the proposed HGWO based soft sensor provides a better result compared to the GWO algorithm.