To more reasonably partition fuzzy spaces during online identification of T-S model, a calculation method on
overlap coefficient between two fuzzy spaces is proposed. In this method, the overlap coefficient can be derived by the
centre distance between two contiguous clusters. In addition, an online T-S model identification algorithm which has higher
identification accuracy can be obtained through the integration of this method, least square(LS) algorithm and recursive
least square(RLS) algorithm. Based on the data of thermal field from a single crystal furnace, the thermal field model is
on-line identified by this identification algorithm. Simulation results show that the single crystal furnace thermal field model
identified by this method has higher precision.