掘进扰动下，煤炭深部开采高地应力会导致深部围岩大范围塑性破坏和强烈动力失稳，且持续时间较长，存在安全隐患，严重威胁着掘进工作面煤矿工人的人身安全。近年来，国内外学者在深部围岩压力变化规律、围岩控制理论和超前支护等方面开展了许多研究，但仍存在许多亟待解决的问题。超前支架的支撑力自适应围岩压力变化可充分利用巷道围岩的自承载能力，避免巷道围岩出现破裂、碎裂等现象。介绍了一种自移式超前支架的结构，分析了该支架支撑力液压控制系统，建立了该控制系统的数学模型，研究了该系统在无控制算法、传统PID控制和神经网络PID控制等情况下的系统性能;根据旗山矿的地质条件建立了围岩-超前支架力学耦合模型，利用FLAC^3D软件仿真拟合得到了巷道围岩压力变化曲线;并以此曲线为输入，模拟研究了在传统PID控制和神经PID控制两种算法时超前支架支撑力对巷道围岩压力变化的自适应变化规律。研究结果表明，与传统PID控制相比，采用神经网络PID控制算法后，支架支撑力控制系统调节时间约为2 s，缩短了16倍，超调量约为6%，动态性能得到了改善;自适应围岩压力跟踪误差为0.005 5，改善了6.8倍，证明了神经网络PID控制策略能够对支架支撑力进行控制且其控制效果比传统PID控制效果有较大优势。

Under excavation disturbance,high crustal stress in deep coal mining will lead to large-scale plastic failure and strong dynamic instability of deep surrounding rock,which lasts for a long time and seriously threatens the personal safety of coal miners at the excavation working face. In recent years,a lot of research have been carried on the variation law of deep surrounding rock pressure,surrounding rock control theory and advance support,but there are still many problems to be solved. In order to avoid the phenomenon of breakage and fragmentation of roadway surrounding rock, the support force of advance bracket should adapt to the change of surrounding rock pressure,so as to make full use of the self-supporting capacity of roadway surrounding rock. The structure of one kind self-forward advance bracket is in- troduced and its hydraulic control system is analyzed,and the mathematical model of the control system is established. The perfor-mance of the control system is studied with no control algorithm,traditional PID control algorithm and neu- ral network PID control algorithm. According to the geological conditions in Qishan mine,the mechanical coupling model of surrounding rock-advance bracket is established,and the curve of roadway surrounding rock pressure is ob- tained by FLAC3D simulation fitting. Taking the curve as the input,the self-adapting change rules of the advance brack- et support force to the roadway surrounding rock pressure are simulated when the traditional PID control algorithm and neural network PID control algorithm are used. The results show that compared with the traditional PID control algo- rithm,the settling time of support force control system with neural network PID control algorithm is about 2 s,which is shortened by 16 times,and the overshoot is about 6% ,and the dynamic performance is improved,and the surrounding rock pressure tracking error is only 0. 005 5,improved by 6. 8 times. The research results show that the advance brack- et control system based on the neural network PID control strategy can control the support force of the advance bracket adaptively with the pressure of the surrounding rock and its control effect is more advantageous than that with the tradi- tional PID control strategy.