Classical clustering methods tend to be less effective in such situation where the data are insufficient or impure. Therefore, two knowledge transfer mechanisms for fuzzy partition clustering are devised in terms of historical cluster centers and fuzzy memberships regarding historical class centers respectively. And combining these two transfer mechanisms with the classical maximum entropy clustering(MEC) approach, the particular knowledge transfer based maximum entropy clustering(KT-MEC) algorithm is proposed. The major merits of KT-MEC lie in following three aspects. Benefiting from the auxiliary guidance of historical knowledge, the clustering effectiveness and practicability of KT-MEC are enhanced distinctly. As the couple of built-in transfer mechanisms both don’t expose the raw data in the source domain, KT-MEC is of good capability of privacy protection for the source domain. Owing to the “searching for best parameters + validity indices”mechanism, the clustering effectiveness of KT-MEC is not worse than that of MEC in theory, which avoids reliably the negative transfer risk.