The Research Of Multi-BBO Algorithm

Intelligent optimization algorithm,also known as modern heuristic algorithm,is a versatile algorithm.It is suitable for parallel processing to find the global optimization.Intelligent optimization algorithm performs well in many engineering applications.In recent years,biogeography optimization algorithm attracts the attention of many scholars in the field of intelligent optimization algorithms and engineering research at home and abroad.Biogeography optimization algorithm updates habitats continuously through migration,mutation,and de-duplication.Among them,the migration operator can exchange information between different habitats,which increases the development ability of the algorithm;the mutation operator can change the state of the current habitat,which improves the exploration ability of the algorithm;the de-repetitive operator is used to remove repeated habitats in order to ensure the diversity of solutions.As a parallel search algorithm,BBO has broad application prospects in many complex optimization problems.In this paper,we introduce the research background and significance of biogeography optimization algorithm,and describe its mathematical model and basic process,and summarize the relevant research status at home and abroad.In the view of the problem of BBO algorithm in solving high-dimensional problems,including insufficient adaptability,low efficiency and high volatility,we propose three improvements and apply improved BBO algorithm to solve TSP problems of different scales.The main work of this paper includes:(1)Because the original BBO algorithm's linear migration operator lacks adaptability,we propose three nonlinear migration models,namely Logistic migration model,cubic polynomial migration model and hyperbolic tangent migration models;in order to improve the adaptability of mutation model,we propose three new mutation rate models,namely Beta mutation strategy,Gamma mutation strategy and Poisson mutation strategy.The test results based on 17 typical benchmark functions show that,for the migration model,under the majority of the test functions,the BBO algorithm based on the hyperbolic tangent variant migration model has the best optimization ability,and its solution is closer to the global minimum of the function.For the mutation strategy,under most of the test functions,the Beta mutation strategy,Gamma mutation strategy and Poisson mutation strategy perform very well,they're better than the original mutation model and related improved algorithms(including Cauchy mutation strategy and Gaussian mutation strategy).(2)Aiming at the problem of low efficiency and high volatility in processing high-dimensional optimization problems with BBO algorithm,a multi-group biogeography optimization algorithm Multi-BBO was proposed,which uses K-means clustering to decompose the habitat space into several sub-populations.At first,sub-habitat does optimization respectively.And then,all sub-habitats exchange information with each other.Next,sub-habitats are merged into a complete habitat.Muti-BBO improves the efficiency and stability of the algorithm when dealing with large-scale optimization problems.At the same time,we introduce the idea of simulated annealing into BBO.We set the mutation operation to be a preferred variation.That is,for each habitat's mutation operation,the habitat accepts a worse solution with a certain probability,which helps the algorithm to jump out of the local optimum.(3)The Multi-BBO algorithm is applied to solve traveling salesman problems of different scales.Based on the test data,experimental results show that the Multi-BBO performs well on small datasets,which is better than large datasets.