Abstract: The operation process of a multiproduct oil pipeline not only has typical spatio-temporal characteristics, but also its operation mode needs to be comprehensively characterized by multiple monitoring parameters. However, the existing spatio-temporal pattern analysis methods make it difficult to reveal the comprehensive spatio-temporal characteristics of multiple parameters. Therefore, a tensor decomposition method based on multi-parameter fusion is proposed to extract the multi-parameter spatio-temporal pattern of multiproduct pipeline operation. Firstly, this method realizes group fusion by analyzing the amount of information and correlations of multi-dimensional monitoring parameters of pipeline operation from different analytical perspectives, and then models the fused spatio-temporal data as tensors and uses tensor decomposition and clustering methods to obtain the multi-dimensional spatio-temporal patterns of the data set. Finally, through comparing the changing trend of the original multiple parameters under different patterns, the spatial and temporal law of the operation pattern is further found. Based on this method, a visualization system MPVis is designed to extract and visualize the comprehensive spatial and temporal pattern of multi-parameter representation from different analytical perspectives. The results from case studies on real-world data show that the method provides a new thinking way for the later in-depth analysis of multiproduct oil pipeline data.