In view of the good properties of the extreme learning machine(ELM) feature mapping, a novel technique of constructing to condensed hidden feature space is proposed by combining principal component analysis(PCA) with ELM feature mapping. The cascaded hidden space fuzzy C means clustering algorithm is proposed to improve the learning ability of the non-linear data. Experimental results show that the proposed algorithm is not only efficient and robust for high-dimension data, but also insensitive to the fuzzy index of fuzzy clustering algorithms.