The Research And Application Of Hybrid Swarm Intelligence Optimization

As optimization problems exist widely in scientific research and engineering practice.Swarm intelligence optimization is the latest branch of optimization algorithm,and it is also the most popular development direction.Swarm intelligence optimization is a kind of stochastic search algorithm,which based on the imitation of the group behavior among natural creatures usch as cooperation and information sharing.Compared with the traditional optimization,it has the advantages of simple structure,easy implementation and so on.Different swarm intelligence optimization imitates different biological behaviors,therefore they have different characteristics and applicable problems.However,a single swarm intelligence optimization has its limitations,such as low search accuracy,slow convergence speed,being affected by parameters greatly and falling into local optimum easily.Generally,creating new hybrid swarm intelligence optimizations by combining different swarm intelligence optimizations is an effective method to improve the performance of the algorithm,hence it has important research significance.The main research and innovation are included in the following several aspects.Firstly,a F-APSO(Follower Bee Search Based Adaptive Particle Swarm Optimization)is proposed for solving single-objective numerical optimization problem.An APSO(Adaptive Particle Swarm Optimization)is presented on the basis of trajectory analysis of the traditional PSO,accordingly the performance in solving unimodal problems is enhanced.Then,a stability analysis method for particle swarm optimization with adaptive parameter is presented,and the stability parameter conditions of the APSO are obtained.Furthermore,the F-APSO is carried out by introducing the follower bee search with reasonable exploiting ability.And the stability parameter conditions of the APSO is extended to the F-APSO.Simulation results show that the F-APSO performs well in solution quality and time consumption for solving single-objective numerical optimization problems.Moreover,the F-APSO is applied to solve the scheduling optimization problem of mine production.Compared with the original production scheme,the optimized scheme can make more profits under different iron concentrate prices.Secondly,a F-AMOPSO(Follower Bee Search Based Adaptive Multi?objective Particle Swarm Optimization)is proposed for solving multi-objective numerical optimization problem.The solution of the multi-objective optimization problem is not the only optimal solution,but a set of non-dominated solutions.Consequently,the optimization for solving single-objective problem can not be directly used to solve multi-objective problem,it is necessary to create multi-objective optimization by reconstructing single-objective optimization aiming at the characteristics of multi-objective problem.In order to design the F-AMOPSO,the updating mechanism of personal best solution is modified,and the selection mechanism and the search range adjustment method of follower bee based on the distance of non-dominated solutions are proposed.Moreover,the simulation results show that the F-AMOPSO has good convergence and distribution.In addition,the method is applied into multi-objective scheduling optimization problem of mine production and excellent practical result is achieved.Finally,a C-PSO-ACO(Chaos Particle Swarm Optimization based Ant Colony Optimization)is proposed aiming at a typical combinatorial optimization problem——TSP(Traveling Salesman Problem).A pheromone update method which combines the global asynchronous feature and elitist strategy and a strategy that reducing the iterations of ACO invoked by PSO are introduced to eliminate the large time-consuming caused by the repeated running of ACO.But few iteration steps of ACO results in imperfect convergence and too strong randomness.Therefore,the quality of solution is inferior.In order to restrict the randomness and enhance the quality of the solution,two chaos sequences with controllable randomness are applied in both classical particle swarm optimization and classical ant colony optimization.Consequently,the simulation results of several TSPs show that the C-PSO-ACO has a better performance in search speed and solution quality compared with other algorithms.Moreover,the algorithm is applied to solve the path planning problem of landfill inspection robot and a shorter path is obtained.