Research On Regional GPS Height Fitting Model Based On Combined Intelligent Algorithm

With the advent of the era of big data,space geodetic technology has developed rapidly,providing the necessary data sources and reference for the field of geographic information.As an important data for the expression and transmission of spatial information,elevation provides a basic guarantee for the practical application of scientific research and engineering.GPS technology with fast and high precision has become an important way to obtain geographic data,but the GPS observation method directly obtains the height of the ground relative to the reference ellipsoid,rather than the normal height actually used by the engineering.Therefore,in order to make full use of the effective information of GPS observations,it is not necessary to construct a high-precision conversion model to achieve the conversion between different elevations.The core step is to solve the elevation anomaly.In order to obtain the optimal elevation anomaly value of the local measurement area,this paper used the swarm intelligence algorithm to optimize the model parameters based on the multi-faceted function,and proposed an improved GPS elevation fitting method based on the swarm intelligence algorithm to construct a regional elevation fitting model suitable for terrain fluctuations and complex terrain conditions.The main work is as follows:1.The relationship between different elevations was systematically discussed.The conversion precision and characteristics of astronomical geodetic method,geoid model method,gravity measurement method and numerical simulation method are compared and analyzed.In this paper,the application scope and different advantages and disadvantages of different fitting methods are analyzed from the linear fitting model,the planar fitting model and the improved combined model,the limitations of the common fitting model and the limitations of the improved method are discussed.2.Considering that the observed values may contain gross errors,a robust estimation was embedded in the multi-faceted function for model optimization,and the modeling data with gross error was processed by the fitting method before and after the embedded robust estimation.The comparative analysis found that the improved multi-face function fitting method helps to weaken the influence of the gross error on the accuracy of the model.3.Aiming at the problem that the two important parameters of the kernel function and the smoothing factor were difficult to select in the multi-face function,an ant colony algorithm and a genetic algorithm were proposed to obtain the combined fitting method of the central node and the smoothing factor respectively.The ant colony algorithm is used to select and optimize the central node,and the optimal center node is compared with the fitting result of the uniform grid alone.The results show that the optimized center node was more conducive to the establishment of high-precision fitting model.The effects of different crossover probabilities on the computational speed and convergence of genetic algorithms were compared and analyzed.The optimal smoothing factor of the experiment was determined and determined to be 0.952.At this time,the accuracy of the fitted model was higher than that of other values.4.In order to further demonstrate the effectiveness of the swarm intelligence algorithm on the improved model,this paper used the particle swarm optimization algorithm to obtain the model parameters,and compared the same check data with the ant colony algorithm.It is found that different groups of intelligent algorithms have better effects on the optimization of the central node.However,the convergence of the particle swarm optimization algorithm is slightly higher than that of the ant colony algorithm,and the obtained central node can express the geomorphic features,so the modeling accuracy is slightly better than the ant colony algorithm improved model.The experimental results show that compared with the multi-faceted function fitting method before the improvement,the improved GPS elevation fitting model based on the combined intelligent algorithm has improved the accuracy,which shows that the improved model is effective and feasible for improving the accuracy.The normal elevation data provides theoretical guidance.