To evaluate the serve status of bridge operation precisely，based on the bridge 3D point data model ob? tained at different times，the detection and tracking method of the bridge geometrical profile based on point cloud data and engineering knowledge is established. Firstly，based on the traditional ICP registration，combined with engi? neering professional knowledge，a point cloud registration algorithm that only uses relatively fixed points to partici? pate in the registration process is proposed. This algorithm is used as the main body to construct a bridge spatial de? formation identification method including the point cloud segmentation，registration，deformation identification and result verification. At last，the set of methods is applied to an engineering example to efficiently obtain the change of geometrical profile of each component of the bridge within one year，and the structural engineering knowledge is used to verify the reliability of the identification results. The results show that the proposed point cloud registration al? gorithm based on engineering knowledge can efficiently realize the precise registration of two point clouds. The change detection method of the bridge geometrical profile based on this algorithm can quickly and accurately identify the spatial deformation of bridge components（e.g. bending and torsion of girders and lateral bending of piers）in a period of time. This proposed method can provide more technical means and practical experience for the intelligent non-destructive testing of bridges.