Research And Application Of The Ant Colony Optimization

Ant Colony Algorithm (ACA) is one kind of bionics optimization algorithms, which simulates the kingdom of the ant community to look for food in the recent development. Though it has demonstrated excellent performance and tremendous development potential to solve many complex combinatorial optimization problems, it is not mature enough to solve the practical problems so that there still exits much development space. After studying a great amount of relevant literatures, some possibility methods of the ACA improvements are proposed in this paper to correct some flaws in solving the practical combinatorial optimization problems with current ACA.The main research works are that the algorithm model is improved on the basis of the experimental analysis of ACA, there are primarily three faces: first, make experimental analysis about the series of algorithm parameter in order to find the reasonable parameters range and combination principle ; second, on the term of updating pheromone mechanism, the paper firstly introduces the concept of "transcendental factor" to better consider the previous "experience "so that avoid unnecessary searching; third , Hopfield Neural Networks (HNN) is combined with ACA to solve the TSP . When solving the practical TSP, the map composed by all cities is at first abstracted as undirected complete graph, then the TSP is converted to one Hamilton cycle of the undirected complete graph without cross-path by pretreatment, a combination of the ant colony optimization and the HNN model of artificial neural network is used to solve the problem. In the paper, the improved ACA is used to seek a better combination of all parameters of HNN, and then TSP is solved by the HNN which has been trained by ACA.At last, the experimental results show the high efficiency of the improved ACA and feasibility of the proposed method based on HNN-ACA by solving the EU51TSP and Att48TSP of the benchmark storeroom.