Cubature kalman optimizer: A novel metaheuristic algorithm for solving numerical optimization problems.

Abstract

This study introduces a new single-agent metaheuristic algorithm, named cubature Kalman optimizer (CKO). The CKO is inspired by the estimation ability of the cubature Kalman filter (CKF). In control system, the CKF algorithm is used to estimate the true value of a hidden quantity from an observation signal that contain an uncertainty. As an optimizer, the CKO agent works as individual CKF to estimate an optimal or a near-optimal solution. The agent performs four main tasks: solution prediction, measurement prediction, and solution update phases, which are adopted from the CKF. The proposed CKO is validated on CEC 2014 test suite on 30 benchmark functions. To further validate the performance, the proposed CKO is compared with well-known algorithms, including single-agent finite impulse response optimizer (SAFIRO), single-solution simulated Kalman filter (ssSKF), simulated Kalman filter (SKF), asynchronous simulated Kalman filter (ASKF), particle swarm optimization algorithm (PSO), genetic algorithm (GA), grey wolf optimization algorithm (GWO), and black hole algorithm (BH). Friedman's test for multiple algorithm comparison with 5% of significant level shows that the CKO offers better performance than the benchmark algorithms.