Ant Colony Algorithm With Semi-dynamic Candidates List

Ant colony algorithm is a new evolutionary computation method which is inspired by the real ant colony's foraging mechanism. Researchers have improved the ant system according to different strategies and developed many different versions of the ant colony algorithm and successfully applied to optimization questions. At present, many domestic and foreign researchers and research institutions have carried out on the ant colony algorithm theory and application research. The VUB of Brussels, Belgium IR1DIA laboratory, Switzerland IDSIA laboratory are currently active ant research institutions of colony algorithm. Overall, the ant colony algorithm has become the hot spot in the field of intelligence computation.This paper based on the ACS nearest neighbor candidate list algorithm. Use the thought of the process of limited generation, design a semi-dynamic of mechanisms to generate a candidates list, making ant colony to solve the problem with a certain degree of memory and the ability to identify surroundings. Using this method, improve the AS algorithm, optimize the search process. At the same time, through the simulation results that contrast with the candidates list of semi-dynamic algorithm AS and AS algorithm in the calculation of basic TSP problem the results verify validity the optimization of algorithm. Through the analysis of the data, using the A -branching factor analysis the different behaviors of ant system algorithm and improved algorithm in different parameters sets. verified the optimization of the search behavior through the use of the candidate list.