Prediction of Surrounding Rock Deformation of Cavern in Main Powerhouse of Pump Storage Power Station Based on Combination of Multiple Influencing Factors
In order to ensure the safe state of the underground cavern surrounding rock environment,a cavern sur-rounding rock deformation prediction method based on the joint kernel-extreme learning machine method(KELM)with multiple influencing factorswas proposed on the basis of the Variable Modal Decomposition(VMD)method to decompose the original data and the Particle Swarm Optimization(PSO)algorithm to improve the prediction accuracy.The method firstly used the VMD method to decompose the monitored displacements into trend term displacements affected by tren-ding factors and cycle term displacement affected by the cyclical factors,and removed the interference terms of the influ-encing factors.Secondly,the evolutionary state and influencing factors were taken as the input data of the PSO-KELM to predict the trend term or cycle term displacements corresponding to the influencing factors.Finally,the two types of dis-placementswere superimposed,and the accuracy of the KELM method combining the multi-influencing factors to other prediction methods was compared.The validation results of the measured data of a pumped storage power station project show that the difference between the prediction results and RRMSE of the original displacement is only 0.76%,and the R of the two is 0.986.The above prediction method has higher prediction accuracy,and it can be used as a reference for the prediction of the surrounding rock deformation of the same kind of projects.
万方数据
维普
