Sudden Grid-Clustering Method Based on Improved Multi-Variety Ant Algorithm

Spatial clustering is important to the application of artificial intelligence in spatial information system. After introducing clustering´s basic concept, some shortcomings of current clustering algorithms were analyzed, and the steps of improving on them were given as the following. First, a way of transforming vector data to grid was presented, and a formula of calculating grid-unit´s value was defined. Second, three type ants and their mutual-transformation modes were defined, which were used to regulate ants´ action so as to improve algorithm´s intelligence and agility. Third, based on the grid technique and multi-variety ant algorithm, a new clustering method came into being, and its whole process was expounded. Finally, an example was illustrated for clarifying the theory of this method, and its advantages were pointed out. Analysis indicates that compared with existent clustering algorithms, the method created in this paper satisfies the requirements of speed, intelligence and independence in spatial data clustering