A Study On Parallel Multiobjective Intelligent Optimization Algorithms And Applications

Optimization problems are almost ubiquitous in daily life,and generally involve multiple objectives that conflict with each other.With the development of multiobjective optimization,the searching space becomes larger apparently,which leads to a huge increase in difficulty of finding optimal solutions.A combination of optimizing mechanisms of intelligent optimization algorithms and high performance parallel computing is a reliable solution to multiobjective optimization problems.Based on strategies of NRTPA,a novel parallel nondominated ranking team process algorithm(PNRTPA)is proposed with the introduction of adaptive rules.PNRTPA takes advantage of two-group behavior-division searching and nondominated ranking strategies,to achieve the approximation to the solutions and diversity on the real Pareto front.With the test of benchmark functions and multiobjective metrics,numerical results verify that PNRTPA and APNRTPA possess the properties of fast convergence,uniform distribution.Meanwhile,the performance of algorithm stability improves due to adaptive rules.APNRTPA is employed to optimize a linear antenna array,which aims to suppress the side lobe level and design the angle of null.The process of establishing objective functions is introduced in detail by utilizing the symmetrical structure and radiation characteristic of antenna array to reduce the searching space significantly.A group of solution is found and the two objectives are optimized simultaneously,and hence it indicates that APNRTPA is applicable to solve practical electromagnetic optimization problems.