An improved ant clustering analysis algorithm is proposed，which modifies the similarity function of the Lumer–Faieta(LF) groups，and adds parameters to adjust the adaptive strategy，controls the random movement of ants by using the short-term memory and grid of the local pheromone distribution，combining with dynamic changing of ants speed，increasing radius，forcing down. Comparison experiments are performed by different algorithms with testing data. The results indicate that the proposed algorithm shows a high stability and a high accuracy.