Improved BP Neural Network Algorithm for GPS Height Conversion

GPS height conversion is the key problem in survey engineering. To resolve it, researchers supposed some mathematical function to fit the surface of normal height such as plain surface, quadratic surface fitting and so on,however, which have model error. It should reduce the accuracy of the fitting result. In order to eliminate and reduce the model errors, this paper adopted BP neural network to convert the different height in the paper. Designed and programmed Matlab code were used to implement the supposed and compared results with other fitting methods, then it was convinced that BP neural would be accuracy, dependability and stabilization.