Metaheuristic‐based approach for state and process parameter prediction using hybrid grey wolf optimization

Abstract

Metaheuristic-based optimization algorithms can be used to solve the complexities in estimating the parameters and states of a complex nonlinear process. In this work, a hybridized version of grey wolf optimizer is employed to solve the intricacies of a complex linear process using hybrid grey wolf optimization. The proposed algorithm is executed simultaneously in 2 steps: (a) predicting the parameters and states using grey wolf optimizer and (b) updating the predicted parameters using static modified Kalman-Bucy mechanism. The significance of the hybridization is that it improves the convergence rate towards an optimal solution and explores global optimum effectively over a complex search space. The performance of the proposed algorithm is tested through 10 typical benchmark functions and is compared with the parent algorithm, grey wolf optimizer, and conventional particle swarm optimization. Furthermore, the findings of the hybrid grey wolf optimization are compared with the outcomes of the conventional algorithms to establish the effectiveness of the proposed hybrid grey wolf optimization in process parameter estimation.

Real-time filtering with sparse variations for head motion in magnetic resonance imaging

“…However, these methods would have to incorporate the unknown sparse innovations in the cost function used to compare solutions to the parameter estimation problem. Related heuristics like parallel particle swarming, grey wolf pack-based optimization, or humpback whalebased spiraling optimization over the entire search space, can attempt to solve for these parameters directly, or in combination with conventional Kalman-based techniques [43,44]. Another such heuristic, the cuckoo search uses a heavy-tailed step length to explore the parameter space more efficiently [45].…”