Research On Prediction Model Of Railway GSM-R Field Strength Coverage

Global system for mobile communications-railway(GSM-R)is a wireless communication system widely used in railway construction in China.To make the GSM-R wireless communication system play a better role in railway applications,it is necessary to arrange the network coverage of railway lines reasonably in the early engineering planning.To achieve this goal,it is necessary to use the relevant wireless communication propagation model to predict the reception level value more accurately.To get a better prediction model,based on the analysis of different GSM-R application scenarios,this thesis uses data mining technology to research and analyze the GSM-R wireless communication propagation model by searching and extracting useful information from a large number of data and finally establishes a GSM-R wireless field strength coverage prediction model suitable for different railway scenarios.An accurate prediction model can not only provide guidance for the network planning in the early stage,but also save manpower and capital costs for the railway construction in the later stage.Firstly,this thesis briefly introduces the propagation characteristics of wireless communication waves in different environments,as well as network communication in high-speed railway environments.In addition,this thesis summarizes the commonly used wireless communication propagation models and provides a detailed introduction to the GSM-R system structure and its functions in railway communication.Based on the impact of different environmental factors,the GSM-R application scenarios are divided into three scenarios:roadbed,bridge,and tunnel,and the propagation models in these three scenarios are modified.And sorting out the obtained data and selecting features.Secondly,due to the complexity of the railway environment,to improve the accuracy of GSM-R field strength coverage prediction,this thesis constructs corresponding prediction models according to the characteristics of different scenarios.For tunnel application scenarios,this thesis introduces the particle swarm optimization(PSO)algorithm to optimize the Radial Basis Function Neural Network(RBFNN)to establish the PSO-RBF propagation model and then establishes GSM-R field strength coverage prediction model based on PSO-RBF neural network algorithm combined with GSM-R field strength.Aiming at the shortcomings of the PSO algorithm,such as premature convergence and easy falling into local optimum,this thesis improves the PSO algorithm by using the method of dynamic convergence of inertia weight and tests the performance of the improved algorithm by using test functions.The results show that the overall search ability of the algorithm has been improved.For bridge application scenarios and subgrade application scenarios,this thesis proposes the Grey Wolf Optimization(GWO)algorithm to optimize the Support Vector Regression(SVR)to establish a GWO-SVR propagation model and then establishes a GSM-R field strength coverage prediction model based on the GWO-SVR algorithm combined with GSM-R field strength.To solve the problem that the searching ability of the GWO algorithm is not balanced when searching locally and globally,the convergence factor a in the algorithm is improved from linear decreasing to nonlinear decreasing,and the performance of the improved algorithm is also tested by using the test function.The results show that the overall searching ability of the algorithm has been improved.Finally,the established PSO-RBF prediction model and GWO-SVR prediction model are used to predict the coverage of GSM-R field strength in different application scenarios,and the established prediction models are evaluated by three evaluation indexes:Total Sum of Squares(TSS),Residual Sum of Squares(RSS)and Coefficient of Determination(R~2).Finally,the GSM-R field strength coverage prediction model suitable for different railway scenarios is obtained,which provides a certain guiding significance for the construction of railway engineering projects.