Multi Objective Barnacle Mating Optimization for Control Design of a Pendulum System

Abstract

This paper presents a MultiObjective Barnacle Mating Optimization (MOBMO) and its application to optimize controller parameters for an inverted pendulum system. The algorithm is an extended version of a single-objective Barnacle Mating Optimization (BMO). In terms of solving a complex problem that has two conflicting objectives, a multiobjective type BMO is needed. Therefore, in the proposed MOBMO, nondominated sorting and crowding distance approaches are incorporated into BMO as a technique to formulate the multiobjective algorithm. The proposed algorithm is tested on various multiobjective benchmark functions. Its performance in terms of accuracy and diversity attainment to find a theoretical pareto front solution is analyzed. Moreover the proposed MOBMO is applied to optimize control parameters for PD controls of a pendulum system. The performance of the proposed MOBMO is compared with Multiobjective Water Cycle Algorithm (MOWCA). Result of the benchmark functions test shows that the proposed algorithm has attained a higher accuracy and a competitive diversity in locating the theoretical front solution. For its application to optimize PD control parameters, both MOWCA and MOBMO have successfully attained a good pareto front solution and controlled the pendulum sufficiently good. Overall performance, the proposed MOBMO has outperformed MOWCA for accuracy attainment and achieved the same level of diversity performance.