Adaptive Immune Multiobjective Algorithm for LSSVM-based NARMAX model

The LSSVM-based NARMAX model and the identification of the model are studied in the paper. By extending the immune cell subset theory and clone selection principle, a novel adaptive immune multiobjective algorithm (AIMA) is proposed. It is shown that, according to the variation of subsets in the memory set, the AIMA can produce different cell subsets adaptively, and thus regulate the related model parameters adaptively to realize the optimization. The simulation result demonstrates that with the parameters identified by the AIMA, the LSSVM-based NARMAX model has a satisfactory accuracy and it can be used to solve dynamic modeling problems.