A fuzzy K-plane clustering algorithm based on double knowledge representations about incremental feature learning and homogeneous style of data(ISF-KPC) is proposed. Before partitioning data samples into different groups, i.e., clusters, the feature augumentation is firstly conducted in an incremental manner based on the original inputs. Since data samples originating from a group share a same homogeneous style, the style information of each group, denoted by a style matrix, will be iteratively determined. By extensive experiments, the proposed ISF-KPC based on the double knowledge representations can obtain comparative clustering performance when compared to the adopted comparative clustering methods, especially it has its best clustering performance on the datasets with clearly homogeneous styles.