Studies on Fuzzy C-Means Based on Ant Colony Algorithm

摘要

A fault identification with fuzzy C-Mean clustering algorithm based on improved ant colony algorithm (ACA) is presented to avoid local optimization in iterative process of fuzzy C-Mean (FCM) clustering algorithm and the difficulty in fault classification. In the algorithm, the problem of fault identification is translated to a constrained optimized clustering problem. Using heuristic search of colony can find good solutions. And according to the information content of cluster center, it could merger surrounding data together to cause clustering identification. The algorithm may identify fuzzy clustering numbers and initial clustering center. It can also prevent data classification from causing some errors. Thus, applying in fault diagnosis shows validity of computing and credibility of identification results.