Research On The New Biogeography-based Optimization Algorithm And Its Application

Biogeography-based optimization(BBO)is a new heuristic algorithm that simulates the geographic distribution and migration of species in ecosystems.It has the characteristics of simple implementation,strong robustness,unique search mechanism,and other intelligent algorithms.In comparison,the BBO algorithm has good convergence and stability,and has attracted more and more attention from experts and scholars.As a new intelligent optimization algorithm,although there has been a lot of research on the perfection of BBO algorithm theory,algorithm improvement and practical application of engineering problems,there are still deficiencies.Therefore,on the basis of these existing studies,in order to further improve the search and development capabilities of the basic BBO algorithm,a new hybrid differential evolution biogeography optimization algorithm and improved biogeography optimization based on local search and non-uniform mutation are studied algorithm,and did some exploration work on the practical application of the proposed algorithm.The specific research contents are as follows:Firstly,HDBBO was studied in the paper.Aiming at the problem of slow convergence of differential evolution with biogeography-based optimization algorithm(DE/BBO)in the early stage of iteration,by effectively combining the utilization of the BBO algorithm with the searchability of the DE algorithm,and using the elite retention mechanism to retain individuals with higher fitness,And introduce an inertia weight strategy to adjust the proportion of mutation operations in the hybrid migration operation to improve the global search ability of the algorithm,while adding small probability disturbances to prevent the algorithm from falling into the local optimal solution as the iteration progresses.The simulation of the characteristic test function and the experiment of the job shop scheduling problem(JSP)show that the proposed method has better optimization performance.Secondly,IMBBO based on local search and non-uniform mutation is proposed.Aiming at the shortcomings of the basic BBO algorithm,which requires a large number of iterations to reach the global optimal,slow convergence,and premature convergence,the linear migration model is replaced by a hyperbolic cotangent model that is closer to the natural law,and then the local search strategy is added to the migration operation of the BBO algorithm in order to enhance the global search ability of the algorithm.At the same time,in order to prevent the algorithm from falling into the local optimal solution at the later stage of the iteration,the non-uniform mutation operator is used instead of the basic mutation operator to improve the convergence accuracy of the algorithm.Finally,the improved IMBBO algorithm is applied to train multilayer perceptrons(MLPs)to verify the optimized performance of the proposed algorithm.Aiming at the problems of local optimization and initial value sensitivity that are common in the training of multilayer perceptron models,the feature vectors of the input data of the multilayer perceptron will be introduced and optimized.The training results of the two classification problem models of iris and breast cancer will be introduced.It can be seen that the IMBBO algorithm has higher classification accuracy than several comparison algorithms.To sum up,the research on the theoretical improvement of BBO algorithm in this paper plays an important role in promoting its practical engineering application.