Improvement And Applications Of Ant Colony Optimization

Ant Colony Optimization (ACO) is a new-style simulating evolution algorithm. The behavior of real ant colonies foraging for food is simulated and used for solving optimization problems. Many scholars payed much attention to this algorithm as soon as it came forth. In the past ten years, ACO has already been widely used in varies fields such as communications, transportation and artificial intelligence and made better effect.Based on the basic ACO theory, the thorough and systemic researches on improving ACO and the applications of the improved ACO algorithms in TSP problems, Flow-shop scheduling problems and optimization problems with continuous variables were carried on. The main completed research works include:1. Aimed at the disadvantage that ACO can easily fall into local optima, an improved ACO with pheromone mutation operator and local search strategy named PMACO was put forward. And a detailed theoretical research on the global convergence of PMACO is performed. The pheromone mutation operation provides more possibilities of the route selection during the search, for that it extends the searching space. Then its ability of escaping from local optima is improved, so that the stagnation problem can be solved effectively. Local search strategy can strengthen the local search ability around the current-best solution so as to increase the search efficiency.2. Some typical TSP problems from TSPLIB were chosen to test the performance of PMACO. The results show that the stagnation problem has been greatly improved and that the search precision and speed have been enhanced. PMACO is also applied to Flow-shop scheduling problems, and can get a satisfying rate of finding the global optima.3. A divisional mixed ACO with Particle Swarm Optimization operator solving optimization problems with continuous variables named PSACO was proposed. The search space is divided into many small areas, and each area is given a certain pheromone value. According to the state transition rules, the artificial ants move to the next solution which is generated randomly or calculated by Particle Swarm Optimization. Local search strategy is also added into PSACO so that the search speed and precision is enhanced. The experiment results on several typical functions approve the validity of PSACO.In the end, the conclusion is given, and the further research work is pointed out.