Abstract: The monitoring system of pump station buildings becomes increasingly perfect, with a large number of monitoring points. Currently, the monitoring sequence of single point is mainly used in the safety monitoring of pump station building, which cannot reflect the overall state. In view of this, the vertical displacement of pump station was taken as the research object, and a similarity index and corresponding measurement method were put forward to analyze the similarity of monitoring sequences of deformation for pump station quantitatively based on panel data theory and dynamic time warping algorithm. By introducing the spatial incidence matrix, the zoning method of pump station considering the position of measuring points was proposed. Finally, the cluster analysis model of deformation of pump station based on panel data analysis method was established. Then the validity of the proposed model was verified through a case study of a pump station project of the South-to-North Water Diversion Project. The results show the measuring points can be divided into four zones by using the model proposed, and the characteristics of deformation and load are clarified effectively, thus providing a novel method for the safety monitoring of the pump station buildings.