Time Series Prediction Model For Deformation Of Surrounding Rocks Base On Nerual Network

With the development of China’s coal to be deep, Many mines have experienced soft rockengineering problems. As the soft rock deformation by many unknown factors, resulting inphysical equations based on numerical simulation methods are difficult to display quickly andaccurately reflect the surrounding rock deformation. By mechanism analysis of surrounding rockdeformation, The surface displacement have a long period of non-linear rheological law as avisual performance of roadway surrounding rock deformation. Therefore， the use of rockdisplacement field monitoring data by a certain time build some Timing of training samples byrecursive loop. And excellent non-linear neural network to fit the Rheological law of soft rock,The above method to build the model can predict the new displacement of roadway. With a smallnumber of samples of Time series prediction, Decided to use three kinds of neural networks asgenetic algorithms to improve the BP,RBF,GRNN improve the prediction accuracy， andMATLAB have been used to establish Time Series Prediction model. The projects which wasabout baijiao mine soft rock roadway design used that model to predict, and we have a control tothe factors affecting of result in the practical application. The implementation results from thismethod shows that both the forecasting accuracy and efficiency are at satisfactory levels. Themodel will be of great application value in both supporting design of the soft rock roadway androadway maintenance.