Rna-Ga With Whale Search Based Lssvms For Modeling Multivariable Systems

RNA-GA with whale search based LSSVMs is proposed for modeling multivariable systems. Inspired by whale behavior, the RNA-GA with the search adopts whale shrinking encircling and the spiral update strategies to enhance the local search ability, and the randomly search strategy to improve the population diversity. The numerical optimization results with some benchmark functions shows the superiority of RNA-GA with whale search. The RNA-GA with whale search is adopted to optimize the least squares support vector machines (LSSVMs) for modeling the crude oil fractionation process. The simulation results show that the model established by the proposed multivariate modeling method can reflect the characteristics of the real multivariate process.