In order to solve the problem that the traditional monitoring have great uncertainty and the monitoring results are not accurate, the automatic monitoring method of spatial multi-dimensional data in Geographic Information System is studied by orthogonal transformation processing method. In this paper, we visualize the terrain spatial multidimensional data and road spatial multidimensional data of Geographic Information System, and use artificial neural network to map them. Under the condition of keeping the relative distance of data items unchanged in low-dimensional space, the coordinate offset of spatial data in adjacent time is minimized by orthogonal transformation, which helps users to find the differences of monitoring areas at adjacent time, and realizes the automatic monitoring of spatial multi-dimensional data in Geographic Information System. The results show that the monitoring results of the proposed method are most consistent with those of manual statistics, and the F value of the proposed method is always higher than 0.9. It is obvious that the proposed method has high monitoring accuracy and strong feasibility and reliability.