With the development of mobile Internet and information technology, many applications bring with mass uncer-tain data, including finance, military, LBS, medicine, meteorology and so on. However, traditional methods for deterministic data can no longer apply to uncertain data, so that it is necessary to devise novel solution to deal with uncertain data. The possible world model that has been widely adopted in this field will derive a huge number of possible world instances con-taining deterministic tuples. An uncertain data stream is a series of unbounded uncertain tuples that arrive rapidly. The ER-topk query, a typical query in uncertain data stream field, is challenging to be processed efficiently. In this paper, an approx-imate algorithm is proposed to deal with tht issue with low space-and time-complexities with the help of searching optimi-zation. Experimental results evaluate the efficiency and effectiveness of the proposed methods.