Research On Magnetotactic Bacteria Optimization Algorithm

Magnetotactic bacteria optimization algorithm(MBOA)is a new random search algorithm inspired by the magnetotactic characteristics of magnetotactic bacteria(MTB).MBOA has a good ability to use the information of the best individual and shows excellent performance in solving single objective optimization problems.In this paper,further research into the theory of MBOA will be carried out firstly and the convergency will be proven.The theory and the experimental analysis show the effectiveness of the algorithm;Seconedly,aiming at the shortcomings of MBOA,the algorithm is improved based on different ways of interaction energy computing and magnetosomes(MTSs)regulation process.Finally,MBOA is extended to the field of solving constraint optimization problems and solving the robot path planning problems,and the feasibility and effectiveness is discussed.This paper has done the following research:At first,the biological characteristic of magnetotactic bacteria is introduced.The optimization process of the minimal magnetostatic energy can be considered as the process of continuely regulating magnetic moment,making magnetotactic bacteria orient and swim along geomagnetic field lines,that is,when the MBOA obtains the optimal solution,it corresponds to the state that when the moments of all cells are oriented in the geomagnetic field.MBOA obtains the optimal solution by regulating the moments of cells continually by the process of MTSs generation,MTSs regulation and MTSs replacement.Based on the above processes,the population are evolved and the information are exchanged and shared.In theory,the convergence of the algorithm is proven.Then the appropriate parameters are selected through the analysis of different parameter settings.At last,by testing on benchmark function and comparing with the classical al algorithms and some state of the art algotithms,simulation results show the validity of the algorithm.During the evolution process of the magnetic bacteria optimization algorithm,the diversity is rapidly reduced,and it is easy to fall into the local optimum.To further improve the performance of the algorithm,the process of MTSs generation and MTSs regulation will use different methods.Three improved MBOA are put forward.They are Magnetotactic Bacteria Moment Migration Algorithm(MBMMA),the best individual guided differential energry magnetic bacteria optimization algorithm(BIDEMBOA),and magnetic bacteria optimization algorithm based on random-pairwise(MBOA-RP),respectively.These improved algorithms have improved the diversity of solutions and the ability to solve the problems.Finally,the experimental results show that the improved algorithm has some advantages compared with the common algorithm.The best individual guided differential energry magnetic bacteria constraint optimization algorithm(BIDEMBCOA)is put forward for solving constraint optimization problems(COPs).In the process of constraint handling,the ratio of the feasible solution in the population is analyzed,which can be divided into three kinds of situations,that is,the infeasible situation,the semi-feasible situation and the feasible situation.Different search strategies can be used to adapt to different situations.So it can find the feasible region quickly,and the excellent infeasible solution can be fully considered to increase the diversity of the solutions.In the early stage of the optimization process,when all obtained solutions are infeasible,the main purpose is to find the feasible region and increase the ratio of the feasible solution.When there are some feasible solutions in the population,we should not only make full use of the information of the feasible solution,but also consider the information of the better infeasible solution,that is,the information of individuals which the degree of constraint violation is small,so as to enhance the diversity of the solution.It can be considered as an unconstrained optimization problem when in a feasible solution.The experimental results compared with other constraint algorithms show that the proposed algorithm,BIDEMBCOA has some advantages in solving constrained optimization problems.Finally,BIDEMBOA is applied to solving the robot path planning problem.This paper briefly introduces the common path planning methods.In this paper,the grid modeling method is used to represent the running environment of the robot.In the process of searching path,a path correspond to a cell individual,and the probability of the nodes on the path correspond to the value of variables in the individual.The proposed optimization algorithm will choose an optimal or near optimal path in the multiple paths generated by the iterative optimization.The algorithm adopted the path selection strategy and dynamic parameters of MTSs regulation strategy to improve the performance.The experimental results show that the proposed algorithm has some advantages in solving robot path planning problem.