An Improved Ant Colony Algorithm And Its Application In Structural System Relibility Optimization

Inspired by the intelligent behaviors of the biological groups, the researchers put forward the realization of a new model of Artificial Intelligence--Swarm Intelligence. The core of the Swarm Intelligence is many simple groups of individuals through mutual cooperation to achieve a simple function, or complete a task. The principal part in Swarm Intelligence can show autonomy, reaction learning and self-adaptive characteristics in the environment.The study of Swarm Intelligence began with the Ant Colony Algorithm. As a typical form of Swarm Intelligence, Ant Colony Algorithm is based on ant foraging groups in the process of moving along the shortest path to the development of the biological behavior of a group of intelligent optimization methods.Ant Colony Algorithm was introduced by Dorigo M. and his colleagues. As a novel nature-inspired metaheunstie based on the phenomenon of real ants foraging behavior in the early 1990's. Many scholars are attracted to study ACA and in the past ten years the algrithm has been widely applied to the fislds of combinational optimization, network routing, founctional optimization, data mining, and path planning of robot etc, and good effects of application are gained. Now, ACA has be come an independ branch of intelligent computation and has been discussed as a special session in many intenational conferenees.In this paper, an improved Ant Colony Algorithm--two stages and a variety of sub-group ant colony algorithm was put forward.The main achievements of this dissertation include:(1) Improved the basic Ant Colony Algorithm from the following:First, variety kinds ants were introduced, all kinds of ants in accordance with the rules of a given search, accelerating the evolution of solution while maintaining the solutions of diversity.Second, the search strategy adopted in phases. Early stages, enlarge the initial choice of appropriate probability to increase the opportunity of the good paths to be selected, so that a better pheromone path early in the algorithm has been enhanced. In the late stage, based on the early accumulation of pheromone, the choice probability returned to normal, to ensure that the algorithm does not appear stagnation.Third, the evaporation rate of pheromone was set to a function of pheromone concentration. This was more closer to the essence of natural phenomena. the pheromone update in the way of global pheromone update methods.Test results shows that the improved ant colony algorithm has better convergence speed in TSP,and with a smaller relative error.(2) Applied the improved ant colony algorithm in Series and Parallel system reliability optimization of redundancy. Combined with engineering practice, the number and the type of redundant component were both taking into account.(3)Transformed the structure of complex systems into level network, denoted the solution with vectors,which enable the improved ant colony algorithm to apply to resolve reliability optimization of complex systems.Centralized the pheromone on network node as node attraction strength to ants, guided by the node attraction strength, ants searched efficientiy in the solution space. The results show that the impoved ant colony algorithm can quickly search for the optimal solution, the calculation results better than the algorithm listed.(4)Defined the nodes reliability constraint as " the probability of at least one road leads to ". The improved ant colony algorithm was applied to solve engineer network structure topology optimization.Further approximate estimate was proposed by recursive methord. Which not only reduces the computational complexity of the algorithm,but also improves the algorithm efficiency. The results show that the recursive approximation was more effectively than other algorithms.Finally,the work of this dissertation is summarized and the prospective of future research is discussed.