A novel RNA genetic algorithm for parameter estimation of dynamic systems

Inspired by RNA molecular structure and operators, a novel RNA genetic algorithm (NRNA-GA) with RNA encoding and operators is proposed for addressing parameter estimation problems of dynamic systems. It adopts nucleotides based encoding and some RNA molecular operators, such as permutation operator and stem-loop operator, which is different from conventional genetic algorithms (GAs). An adaptive mutation rate is also used to guard against stalling at local peak. In order to overcome the drawbacks of premature convergence of GAs, a type of special fitness function incorporating objective function values with Euclidean spaces distance is introduced, which leads the population to maintain its diversity and the algorithm to jump out of local optima. A simple direct search method is incorporated into the NRNA-GA to improve local search performance. Numerical experiments about benchmark functions and real-world parameter estimation problems in dynamic systems demonstrate the efficiency and effectiveness of the proposed optimization algorithm.