Aiming at the deficiencies of ant colony clustering algorithms in distance measure, convergence rate, similarity clustering, and other aspects, WALF algorithm is presented as an improvement to the LF algorithm. Using weighted hybrid distance as distance measure, and introducing the adaptive mechanism during the process of clustering, ant colonies can adjust the radius of neighborhoods dynamically and merge similar clusters in the process, meeting the clustering requirements while improving the convergence speed. Finally WALF algorithm has been shown through experiments to be better than LF algorithm both in clustering results and efficiency.