Novel Brain Storm Optimization Algorithms And Their Applications

Brain storm optimization algorithm is a kind of human cluster intelligent optimization algorithm based on creative problem-solving process-brainstorming discussion.This meta-heuristic algorithm combines collective learning with individual learning.It can quickly find the optimal solution and avoid prematurity of algorithm through the coordination of local search and global search.It has been widely used in solving complex optimization problems.This paper takes the brain storm optimization algorithm(BSO)as the research object.The principle and shortcomings of the algorithm was analyzed to make continuous and discrete improvements in a targeted manner.the improved algorithms were applied to the following three specific problems:First,build a model for the prediction of civil aviation passenger traffic.Select the ten most influential parameters to set the prediction function through factor correlation analysis and train the model with an optimization algorithm after the end-average processing of the multi-layer perceptron.Due to the problems that BSO algorithm is easy to fall into the local optimum when solving the continuous optimization problem,and its convergence efficiency can be further improved,this paper combined teaching and learning based optimization algorithm(TLBO)to propose an improved BSO algorithm,teaching-learning and brain-storm based optimization algorithm(TBSO),from the perspective of psychology.The human experience in the process of problemsolving was summarized and was converted into formulas.The idea of teaching and learning under the herd effect was introduced to improve the update method of BSO algorithm.To improve the efficiency of TBSO algorithm,the directionality of individual learning was enhanced and the local search ratio in the later stage of the algorithm was expanded through the step size adjustment mechanism based on the self-selection effect.The TBSO algorithm was applied to train the prediction model,and the reliability of the model was verified through precision,convergence and robust analysis.Second,select the grid map environment to establish the path planning model of the mobile robot.Due to the continuous update method,it is easier for BSO algorithm to adopt coordinate point coding when solving the robot path planning problem.However,by doing so,the initial solution success rate is low.The computational cost is also high,and the path is not smooth.Therefore,brain storm target search optimization algorithm(BTSO)was proposed to solve the discrete problem of the grid environment.Under the framework of the traditional BSO algorithm,the greedy mobile search method was used to obtain the better initial solution.The mutation and update process of the algorithm were redefined,and the path was smoothed in both directions.Experiments were carried out on grid maps of different scales.By comparing the moving paths of robots planned by BTSO algorithm,A* algorithm,particle swarm algorithm and ant colony algorithm,it shows that BTSO algorithm has better performance in grid maps of different scales.It can effectively avoid obstacles and stably plan a smooth and short path.Third,the development of logistics drones needs to be promoted urgently under the Coronavirus disease background.Starting from the problem,this paper established a vehicle-drone joint distribution model.Many factors such as transportation time and distance cost were considered,and distribution path planning was conducted in stages.For the discretization problem of such non-grid environment,crossover operator brain storm optimization algorithm(CBSO)was proposed.The population was set up by means of fragment coding,and the position exchange and crossover operators were respectively applied to the individual update process in the stage of intra-group discussion and cross-group discussion to plan the route network of vehicle distribution.Based on the determination of the vehicle path,a transport weighting graph was established,and the Dijkstra algorithm was used to reasonably allocate the work of vehicles and drones to save distribution costs.The actual network data of a company in Shanghai was selected for simulation.The experimental result shows that the CBSO algorithm can plan a feasible transportation network and improve the distribution efficiency when applied to the vehicle-UAV joint distribution problem of a specific map.By building the mathematical models of the above-mentioned different application scenarios,the feasibility and operation efficiency of the improved BSO algorithm in solving corresponding types of problems are discussed.While expanding the application scenarios of the BSO algorithm,its improvement strategies are enriched.It’s beneficial to the application and improvement of the continuity algorithm.The specific problem scenarios selected by the research have certain practical significance.