摘要
Abstract
The borehole stress in Maochang Aluminum Mine was monitored,and a nonlinear dynamic analysis model for the time-series data of borehole stress was constructed.Based on the framework of chaos theory,the time-series data of borehole stress were analyzed using the phase space reconstruction method.The optimal embedding dimension and time delay of the system were determined to be 5 and 1,respectively.The maximum Lyapunov exponent was calculated to be 0.028 5,confirming the chaotic characteristics of the borehole stress system.The phase space reconstruction revealed the evolution pattern of the phase point distances of borehole stress changes in the phase space.A prediction model for borehole stress changes was established using Long Short-Term Memory(LSTM)deep neural networks combined with phase space reconstruction.The study demonstrated that:① The phase space reconstruction technique significantly amplifies the nonlinear characteristics of borehole stress through high-dimensional space mapping,providing a new theoretical perspective for analyzing ground pressure activity patterns;② The established borehole stress prediction model exhibits high prediction accuracy,with minimum and maximum errors of 0.11％and 1.72％,respectively,and an average error of 0.49％;③ The prediction model accurately captures the chaotic characteristics of the stress system.By integrating chaos dynamics theory with deep learning algorithms,the model enables quantitative early warning of mine ground pressure,offering reliable technical support for safe mining operations.关键词
地压分布/相空间重构/LSTM神经网络/安全预警系统Key words
Ground pressure distribution/Phase space reconstruction/LSTM neural network/Safety early warning system分类
矿业与冶金
