摘要
目前物理信息神经网络成为工程力学数值计算的新范式,其可以在无标签样本数据的情况下,通过融入神经网络物理机理进行偏微分方程的逼近求解,使用较小计算代价快速进行求解。阐明了物理信息神经网络(PINN)计算原理及进行煤体黏滑冲击控制方程求解框架,并通过算例与显式有限差分求解方法进行求解结果的对比,证实了其求解结果的有效准确性。以晋华宫矿12号煤层工作面为研究背景,进行了煤体黏滑冲击动力数值模拟,从数值模拟的手段得到了煤体黏滑变形过程出现时间间隔和波动性规律。通过数值模拟结果可知:当采场支承压力集中系数为2时,煤体黏滑位移(形变)的最大值有明显的增加;当采场支承压力集中系数为3时,采场煤体发生典型的黏滑现象,位移出现准周期性的变化,煤体黏滑系统储存弹性能逐渐增大以及出现煤岩黏滑过程的强度损伤;动摩擦因数对煤体黏滑动力学特性影响较大,当动摩擦因数变化时,煤体黏滑系统位移场数值变化明显。
At present,physical information neural network has become a new paradigm for numerical calculation of engineering mechanics.It can solve partial differential equations by integrating the physical mechanism of neural network in the case of unlabeled sample data,and solve them quickly with low computational cost.In this paper,the calculation principle of physical information neural network(PINN)and the framework for solving the governing equation of coal stick-slip impact are expounded.By comparing the solution results with the explicit finite dfference method,the effective accuracy of the solution results is verified.Taking the working face of No.12 coal seam in Jinhuagong Mine as the research background,the dynamic numerical simulation of coal stick-slip impact is carried out,and the occurrence time interval and fluctuation rule of coal stick-slip deformation process are obtained by means of numerical simulation.The numerical simulation results show that when the abutment pressure concentration coeficient is 2,the maximum value of coal stick-slip displacement(deformation)increases obviously;when the abutment pressure concentration coefficient is 3,the typical stick-slip phenomenon occurs,and the displacement changes quasi-periodically,the storage elastic energy of coal stick-slip system gradually increases,and the strength damage of coal-rock stickslip process occurs.The dynamic friction factor has a great influence on the coal stick-slip dynamic characteristics.When the dynamic friction factor changes,the displacement field of the coal stick-slip system changes obviously.
作者
王民华
闫永敢
WANG Minhua;YAN Yonggan(Department of Mining Engineering,Shanxi Institute of Energy,Taiyuan 030006,China;College of Mining Engineering,Taiyuan University of Technology,Taiyuan 030024,China)
出处
《煤矿安全》
CAS
北大核心
2023年第7期59-68,共10页
Safety in Coal Mines
基金
山西省高等学校科技创新计划资助项目(2022L603)
山西省基础研究计划资助项目(202103021224334)。
关键词
物理信息神经网络
煤体黏滑
冲击地压
动力学特性
数值模拟
physical information neural network(PINN)
coal body stick-slip
rock burst
dynamic characteristics
numerical simulation
