There are many methods proposed to predict transmembrane protein topology. However, these methods have some shortcomings to correctly predict the boundary of the transmembrane region. In this paper, a new transmembrance prediction based on data fusion technology is proposed. The results of different prediction methods, described as different interval number, can be combined with induced ordered weighted averaging operator to determine the appropriate fuzzy interval. The appropriate fuzzy interval can be determined. The transmembrane regions can be predicted by choosing the threshold. The effectiveness of the transmembrane prediction is verified by experiments.