为了提高瓦斯抽采的安全性、效率性、降低抽采的经济成本，分析了瓦斯抽采系统运行的安全约束条件及效率约束条件；分析了瓦斯抽采系统四大控制任务及瓦斯抽采优化数学模型；根据理论调控策略，提出了瓦斯抽采智能调控的完整流程。以上述调控流程为基础，提出了瓦斯抽采智能调控模型，将瓦斯抽采浓度、瓦斯抽采纯量、瓦斯抽采负压及抽采泵效能比作为被控量，抽采钻孔的阀门开度和抽采泵功率作为控制量，利用简单循环神经网络(simpleRNN)分析处理被控量历史数据的时间变化规律，学习得到被控量随时间变化的理想动态拟合曲线。通过模型预测控制算法(MPC)对控制量进行智能调控，使被控量的实际值无限趋近于其理想动态拟合曲线对应时刻的参考值。利用校正反馈和滚动优化不断增强瓦斯抽采智能调控模型的抗干扰能力，最终提高煤矿瓦斯抽采的安全性和效率。以模拟瓦斯抽采数据为例，完成了算法仿真实验。试验结果表明：瓦斯抽采浓度总体变化趋势从40%～5%随时间递减，瓦斯抽采纯量从总体变化趋势9.0～5.0 m3/min依次按时间递减，通过循环神经网络得到的理想动态拟合曲线具有很好的数据拟合度，能较准确的反映瓦斯抽采浓度数据和瓦斯抽采纯量数据的变化规律；瓦斯抽采负压及抽采泵效能比理想动态拟合曲线能较准确地维持在10～30 kPa和1.3～1.5 m3/(kW·h)，满足瓦斯抽采过程的经济和安全需要；模型预测控制算法对控制量的动态调控能克服环境及非线性因素的干扰达到较好的调控效果，为瓦斯抽采智能调控提供了一定的参考依据。

In order to improve the safety and efficiency of gas extraction and reduce the economic cost of gas extraction,the safety constraints and efficiency constraints of gas extraction system operation are analyzed. The four control tasks of gas drainage system are analyzed and the mathematical model of gas drainage optimization. According to the theoretical control strategy,the complete process of intelligent control of gas extraction is put forward. On the basis of the above regulation process,an intelligent regulation model of gas extraction is proposed,which takes gas extraction concentration,gas extraction pure quantity,gas extraction negative pressure and extraction pump efficiency ratio as the controlled quantities,and the valve opening of extraction drilling hole and extraction pump power as the controlled quantities. The simple RNN is used to analyze and process the time-varying law of the historical data of the controlled quantities,and learn the ideal dynamic fitting curve of the controlled quantities changing with time. The model predictive control algorithm (MPC) is used to intelligently control the controlled variable,so that the actual value of the controlled variable infinitely approaches the reference value at the corresponding time of its ideal dynamic fitting curve. Using correction feedback and rolling optimization,the anti-interference ability of intelligent control model of gas extraction is continuously enhanced,and finally the safety and efficiency of coal mine gas extraction are improved. Taking the simulated gas extraction data as an example,the algorithm simulation experiment is completed. The experimental results show that the overall change trend of gas extraction concentration decreases with time from 40%-5%,and the overall change trend of pure gas extraction quantity decreases with time from 9.0-5.0 m3/min. The ideal dynamic fitting curve obtained by cyclic neural network has a good data fitting degree,which can accurately reflect the change law of gas extraction concentration data and pure gas extraction data. What’s more,the negative pressure of gas extraction and the efficiency of gas extraction pump can be accurately maintained between 10-30 kPa and 1.3-1.5 m3/(kW·h),which meets the economic and safety needs of gas extraction process. The model predictive control algorithm can overcome the interference of environment and nonlinear factors to achieve better control effect，which provides a certain reference for intelligent control of gas extraction.