Deformation Prediction of Dangerous Rock Fissure Based on RBF-Markov Series Model

Dangerous rock is one of the common geologic hazards in mountain areas.How to predict deformation of dangerous rock fissure is of great significance to protect people's lives and property and reduce investment for geological control project.On the basis of gradualness and randomness of fissure deformation over time,this paper combines the radial basis function(RBF) model with the Markov model,and then constructs a new RBF-Markov series model used for fissure deformation prediction.The main process is presented as follows: Historical data input in model are trained by RBF and fitting results are output,and then residual errors between fitting values and measured data are classified and relative errors in future k-steps are calculated by Markov probability transition model.When recent in-situ data are put into series model,the final prediction can be realized by the predicted values of RBF model and residual correction values.As an example,the series model has been applied to the prediction of short-term fissure deformation of dangerous rock C1 in Zhong-Wu gas-delivering pipeline based on the in-situ data,and it not only can reflect the tide of historical fissure development,but also considers the influence of the stochastic fluctuation on deformation prediction.The results show the series model has the advantages of high accuracy,rapid convergence,and perfect short-term prediction,so it provides a new approach for prediction of dangerous rock fissure deformation.