Modified Cuckoo Search Algorithmsand Their Applications In Optimization Problems

Problems in many fields of human social life can be described in optimizationproblems. It has been a hot research spot to solve optimization problems. With therapid development of computational intelligence, many an algorithm based onpopulation was applied to complex optimization problems. Such as particle swarmoptimization, glowworm swarm optimization, ant colony algorithm, bee colonyalgorithm and so on. Applied results demonstrated the excellent performance andgreat development potential of swarm intelligence algorithms.Cuckoo Search(CS) is a new metaheuristic search algorithm proposed by Yang XS and Suash Deb in2009. It derives from simulating the aggressive reproductionstrategy of cuckoos. It has been successfully applied to many fields such asengineering optimization, design optimization and so on because of its simplemodel and few parameters. But its optimal performance and application areas canbe further improved. Such as the improvement of solving precision, convergentrates, local search ability and the expand of application areas. Aiming at this, thisthesis made further research of CS algorithm and its application areas and obtainedfollowing research results:(1) Presented a small-scale and multi-population CS algorithm to cope with theproblem that bigger population will leads to longer iteration time. This increasedthe population diversity and reduced the search time at the same time. Experimentsresults show that the proposed algorithm increased the convergence speed andaccuracy.(2) The small-scale and multi-population CS algorithm and CS algorithm wereapplied to economic control systems. The feasibility of the algorithms weretestified by production and inventory system and optimal allocation problem ofmachines. Then we applied the two algorithms to dynamic input-output model. Itcould find out better answers than that in references. Results show that the optimization ability of small-scale and multi-population CS algorithm is superiorto the original algorithm.(3) The step size of CS algorithm was updated randomly which lead to theresults lack of adaptability. We raised variable step adaptive CS algorithm to copethis problem. Experimental results show faster convergence rate and higheraccuracy.(4) We applied the variable step adaptive CS algorithm was applied to chemicaland biochemical dynamic optimization problems. Experimental results of batchreactor, tubular reactor and bioreactor testify this algorithm is a valid method tosolve chemical and biochemical processes dynamic optimization problems.(5) We proposed a new modified method which was called CS algorithm basedon simplex method to overcome shortcomings such as slow converge rate and lowprecision of CS algorithm. Experimental results show the proposed method canreach higher convergence rate,precision and faster convergence speed than CS.(6) The CS algorithm based on simplex method(SMCS) was applied toconstrained optimization problems. Results of reducer design problem, retractablespring design problem and welded beam design problem manifest that SMCS canfind better solutions than the others in references.