Abstract: The elevation used in the current engineering survey is normal height,while the elevation obtained by GPS surveying is geodetic height,the elevation anomaly between them is existed.In order to deal with the height anomaly value,the method of GPS height fitting method based on the powerful data processing function of Matlab sofware,and combing with the engineering examples,by using the programming function of Matlab software to realize the GPS height anomaly fitting.The results show that:①if the test area is with distributed with planar shape,the quasigeoid characteristics of the test area can be reflected effectively by using the surface fitting method;②the fitting effects of the BP neural network fitting method is good,the fitting efficient precise of the combined fitting method is superior to the one of single fitting methods;③the external and internal coincidence precises of the surface fitting method,curve fitting method,BP neural network fitting method and the combined fitting method can meet the requirements of four geometric level.The study results further indicated that it is feasible and effective to apply Matlab software to conduct GPS hight fitting.

Key words: GPS, BP neural network, Height anomaly, Surface fitting, Curve fitting