Research Of Multi-objective Based On Clonal Selection And Its Application

As the development of the human society and the modern industry, more and moremulti-objective optimization problems arise. The solution of the multi-objective optimizationproblems is not single, but is a collection of solutions. Study how to solve the multi-objectiveoptimization problems has been become one of the research focuses of the academic community,multi-objective optimization algorithm are proposed to solve it in the1960s. Currently some ofmulti-objective optimization algorithms have been used in the projects. This paper focuses on animproved algorithm which is based on the theory of multi-objective algorithms and some classicmulti-objective optimization algorithms. Finally, the algorithm is applied to the wireless sensornetwork (WSN)coverageoptimization.The maincontents areas follows:Firstly, it describes the basic concept of multi-objective optimization problems, andmulti-objective optimization algorithms. And then briefly discusses the research status ofmulti-objective optimization algorithms which are based on evolutionary algorithm or artificialimmune system. Also describes the performance evaluation index of multi-objective optimizationalgorithms, andthetest functions which arefrequently used.Secondly, the proposed multi-objective evolutionary algorithm is based on the concentrationclonal selection. Clonal selection of the algorithm draws biological mechanism, a new clonemethod is expressed as a function of antibodies concentration, the antibodies concentration isassociated with the affinity of antibody-antigens and the affinity among antibodies, in order tolooking for a solution which closes to the true Pareto frontier and evenly distributed. The antibody-antibody affinity of the antibody is associated with the distance between the antibody and theantibody’s concentration, the calculation of the antibody-antigen affinity draws the classicalalgorithm SPEA2’s fitness calculation method. Experimental results show that this algorithm inconvergence and solutions’ uniformity are basically better than other more commonly usedmulti-objective optimization algorithms SPEA2, NSGA-II and NNIA when solving the ZDTseriesfunction.Thirdly, a further improvement is applied on this algorithm. CCSMOA involves the suppression threshold value which is just randomly selected based on some experiments, and herewill deep explore and research about suppression threshold. The first, theoretically considerate theinfluence of the too large and too small suppression threshold, then it is set to different values toverify the theory.As can be seen from the experiment the suppression threshold directly affects thequantity and quality of non-dominated solutions, thereby affecting the convergence of thealgorithm. Subsequently, proposed a way to improve the performance of the algorithm withadaptive suppression threshold. Finally, from the contrast experiment found that the improvedalgorithmA-CCSMOAisbetterthan previousalgorithmCCSMOAon theperformance.Next, the new algorithm CCSMOAis used in the optimization of the wireless sensor networkcoverage and the energy consumption. The contents of this chapter are not only in order tooptimize WSN but also in order to verify the algorithm can solve practical engineering problems.Experimental results show that the coverage of the new algorithm for WSN optimization is thefeasible and efficiency, last compared to the other multi-objective optimization algorithms in theWSN optimization results, the comparative experiments show that the optimization results of theproposedalgorithm isbetterthan SPEA2, NSGA-II andNNIA.