The Study On The Analysis And Estimated Model Of Mining Surface Deformation Based On Chaotic Support Vector Machine Algorithm

If the underground working scope achieves a certain scale, the mining subsidence will affect the scope of the development from the rock mass to the surface, causing ground movement and deformation. The ground movement and deformation can bring a series of primary disasters and secondary disasters. Due to the production and the life of the human are mostly carried out at the surface, so the ground movement and deformation relatively large impact on human. Ground movement and deformation of the system is a complex system, in its constant evolution, it exchanges with the outside material, energy and information, showing a strong non-linear characteristic, so this paper mainly studies on the analysis and the forecast of the gob ground movement and deformation based on the chaos support vector machines algorithm.In this paper, it has summarized and classified the analysis and prediction models, which are the analysis and prediction deterministic models, the analysis and prediction Statistical models and the analysis and prediction Mixed models. It has conducted the research to the model of the gob ground movement and deformation forecast. Then it realized the chaos system's forecast procedure, mainly includes the delay time the selection, the inserting dimension selection, the chaos characteristic quantity computation, forecast based on the GA-SVR algorithm and forecast based on the biggest Lyapunov index and so on. The procedure is verified by typical Lorenz chaotic system. The confirmation result indicated that the procedure may apply in characteristic distinction and forecast of the other chaos system.Then it has studied the discrimination method of the system chaotic characteristics. System chaotic characteristics determine the geometry usually through a single invariant, this paper combines qualitative and quantitative approach to distinguish chaotic, that qualitative discrimination Methods power spectrum analysis method, quantitative method of identification selected GP and the correlation dimension algorithm small data sets method for calculating the maximum Lyapunov index method. It has applied support vector return algorithm to the ground movement and deformation time series forecast using genetic algorithm to obtain the three parameters of the accuracy of support vector machine model, the average absolute error of support vector machine prediction as fitness function, namely nuclear function type and parameter value, loss function type and parameter value, penalty parameter value. Finally it had applied the model to the project. The forecasting result indicated that comparing to estimate model which based on the biggest Lyapunov exponential method, estimate model which based on the chaos GA-SVR algorithm is more suitable to the small sampled data and it has very good pan-ability.