Research On The Prediction Method Of Settlement In High-Speed Railway Considered On Regional Settlement

Effective control of the settlement and uneven settlement is an important guarantee for the high smoothness of high-speed railway line engineering. In regional settlement areas, uneven settlement of the ground not only reduce the track design elevation and change the track gradient, but it also cause line engineering settlement with different degrees. These will have a direct impact on the track smoothness and even endanger the normal operation of the railway.Neural network is an important branch of artificial intelligence in the field because of the good nonlinear approximation. It was applied in-depth in many engineering applications. However, the neural network models’performance was poor because of some items, such as training time too long, easy to access to the local optimizations, network initial parameters hard to define, and so on. Genetic Algorithm is based on the view of Darwin’s theory of evolution in the survival of the fittest point. GA transferred the actual problems that need to solve into evolution processes of the biosphere and ultimately find the optimal solution. The ability of global optimization of GA particularly suited to compensate for neural networks to prevent into local optimum situation.In this paper, the Genetic-Algorithm BP Neural Network (GA-BP) is built by means of the fusion of neural network and genetic algorithms and utilizing the good ability of global optimization of the neural network to nonlinear approximation capability and genetic algorithms. The sedimentation survey data of high-speed railway line engineering can be predicted by applying the GA-BP. The main achievement of the paper as follows:1、GA-BP neural network calculation model is built by utilizing genetic algorithms to optimize the initial weights and threshold of neural network. Compared with traditional BP neural network model, the model is improved markedly in some aspects, especially the convergence rate and fitting precision.2、GA-BP neural network model for settlement prediction of high-speed railway line engineering is established. The model accuracy depends on the sample precision of measured settlement and network topology. The predictive value agrees well with measured values, meanwhile, the model can better reflect subsidence situation of the structures.3、Based on GA-BP neural network, the regional superposition engineering subsidence prediction method is proposed. The impact factor of high-speed railway settlement deformation of line engineering is analyzed and studied. According to monitoring features of regional subsidence area, when the information of datum point which can reflect regional subsidence is introduced in genetic BP neural network, it is found that fitting precision of predictive value is increased50%and predictive value is more close to measured values. Meanwhile it can also better reflect accurately subsidence situation of the measuring points.