Research On GNSS Height Fitting Method Based On ABC-FOA-LSSVM

The Global Navigation Satellite System(GNSS)is an advanced navigation and positioning technology that has risen synchronously with the development of modern science and technology.It has all-weather and high-precision characteristics.Now GNSS technology has become an important means of quickly collecting geographic data,but GNSS observation technology directly obtains It is the earth height relative to the reference ellipsoid,not the normal height required in actual engineering.In order to use GNSS technology to replace the traditional leveling technology,constructing a high-precision conversion model has always been the focus of research on surveying and mapping.The key to different elevation conversions lies in the solution of elevation anomalies.In order to obtain the precise elevation anomalies in the local area,this paper proposes to use the bee colony-fruit fly hybrid algorithm to select the best parameters for the least squares support vector machine fitting method to complete the establishment of the GNSS elevation fitting model.The main work content is as follows:1.Briefly explain the definition of elevation and the conversion between different elevations,and list the conventional methods and characteristics of determining the geoid.The respective scope,advantages and disadvantages of the curve fitting method,surface fitting method and BP neural network were analyzed,and the deficiencies of the common fitting methods and the limitations of the improved methods were discussed.2.In view of the possible gross errors in the GNSS observation data,it is proposed to introduce robust estimation in the least squares support vector machine to carry out the height fitting,and compare with the least square support vector machine fitting method without embedded robust estimation The observation data with gross error in the same group was processed.The comparison and analysis found that the improved least squares support vector machine fitting method can reduce the interference of gross error on the model accuracy.3.Considering the difficulty of selecting kernel parameters and regularization parameters in the least squares support vector machine,it is proposed to introduce the bee colony algorithm and the fruit fly algorithm into the least squares support vector machine respectively,and optimize the parameters through the biological intelligent algorithm To construct a GNSS elevation fitting model.By comparing and analyzing different intelligent algorithms to improve the accuracy and stability of the fitting method modeling,it is found that the bee colony algorithm has the characteristics of strong global optimization ability,and the fruit fly algorithm has the advantages of simple calculation process and easy adjustment.4.In order to give full play to the advantages of the intelligent algorithm,this paper uses a bee colony-fruit fly hybrid optimization algorithm to improve the least square support vector machine to build a fitting model,and compares the data of the same checkpoint with the calculation result of a single intelligent algorithm.It is found that the stability and accuracy of the combined algorithm are higher than the single intelligent algorithm,so the combined intelligent algorithm is more suitable for the optimization of the fitted model parameters.The experimental results show that,compared with the conventional least squares support vector machine fitting method,the improved fitting method based on the combined biointelligence algorithm effectively improves the accuracy of the GNSS elevation fitting model,indicating that the improved fitting method can improve the accuracy of the model.The reality and feasibility provide a certain reference for the acquisition of normal high data in the future measurement work.