Inspired by biological immunity mechanism, a novel immune network model named as UOFC-AINet was proposed to specifically perform unsupervised optimal fuzzy clustering. The bottom layer of UOFC-AINet generated optimal centroids of clusters with given cluster number and network parameters, which were controlled by the top layer of UOFC-AINet. Unlike aiNet immune network for data analysis, each antibody in the UOFC-AINet immune network was encoded by a possible solution and optimal antibodies in the network were evolved according to objective function of fuzzy clustering. Based on the clone, mutation, network suppression and influx of new cells, the UOFC-AINet network is capable of maintaining local optima solutions, exploring the global optima and dynamically set number of clusters and parameters of the immune network. The algorithm was described theoretically and compared with similar approaches experimentally. The results of experiments were evaluated with validity measures and visualized by PCA and fuzzy Sammon mapping.