Research And Application Of The Regional Similar Geoidal Surface Models

We get the high-precision geodetic height after the processing of GPS observing data. The geodetic height through the transformation models converts our actual common normal height, which have been difficult to determine, because the quasi- geoid is an irregular curved surface. GPS observation compared with traditional leveling elevation is fast and convenient, Surveying and mapping workers want to the geodetic height into the normal height, so it need to learn out the elevation abnormal and the relationship between the geodetic and the normal height。At present, many domestic and foreign scholars are studying the problem of GPS elevation abnormal fitting. By generalizing and analysis we can put the methods used are divided into three types:Gravity method, geometry analytical method (mathematical model method), the comprehensive method (gravity method integrate geometry analytical method).This paper mainly analyzes using the geometry analytical method to fit the geoidal surface, through the study with the geometry analytical method, we draw some conclusions:1. After it analyzes and summaries the actual methods of the regional similar geoidal surface, then classifies the methods in detail.2. This paper establishes a lot of geometry analytical models, through the examples analyse to quadric surface and Shepard surface of the simple model and mixable fitting model and weighted model of the comprehensive models, it comes to a result that the precision of the comprehensive models are better than the simple models.3. Through the examples, the paper analyzes the determination to the connection weight of BP Neural network. And through the examples draw that fitting accuracy of the BP neural network is superior to the traditional and single geometry model, and the traditional synthesis model fitting accuracy is quite.4. to the BP neural network of fitting method, find out the defects of the BP neural network, Put forward to solve the defects of the problems, Choose adaptive genetic algorithm to do optimize network connection weights and threshold, But the results of discovery optimization effect is not obvious, According to this phenomenon, make mathematics analysis, find out the reasons.