Determining Reconstruction Variables and Parameters on Improved Multi-Objective Immune Algorithm

Abstract

Aimed at multi-dimension of evaluating reconstruction effect and the correlation between reconstruction variables and reconstruction parameters, improved multi-objective immune algorithm (IMOIA) was proposed to simultaneously determine reconstruction variables and parameters in multivariate time series phase space. In IMOIA, antibody coding was binary strings cascade of multi-parameter, antibody evaluating was three functions, and a new population evaluating S - is presented. Immune operators were designed: overall clone operator to improve speed, mutation operator with multiple changing probability on gene loci and S enhanced local and global search capability to ensure algorithm convergence, Clonal Selection operator on Rank and Crowding-distance to ensure solutions uniformity, repetitive antibodies determination and deletion operator to ensure solutions diversity. Use our method to two benchmarking chaos system, simulation results showed that our method can simultaneously solved multiple combinations of optimal reconstruction variables and parameters and had obvious superiority to other intelligent optimization algorithms and determining reconstruction parameter methods.