结合深度确定性策略梯度算法和分级学习的采煤机调高控制策略  

Combining deep deterministic strategy gradient algorithm and hierarchical learning for coal mining machine height adjustment control strategy

作　　者：陈鸿垚 胥良 李高行 CHEN Hongyao;XU Liang;LI Gaohang(School of Electrical and Control Engineering,Heilongjiang University of Science and Technology,Harbin 150022,China)

机构地区：[1]黑龙江科技大学电气与控制工程学院,哈尔滨150022

出　　处：《黑龙江电力》2024年第4期351-359,共9页Heilongjiang Electric Power

摘　　要：针对采煤机截割行进过程中面临大惯性、高负荷以及复杂多变的工况,导致调高响应速度慢、路径跟踪性差以及滚筒和机身振动等问题,提出一种基于深度确定性策略梯度算法的采煤机调高控制策略。通过拟合示范刀的采样位置为截割路径,作为调高控制系统的环境输入,结合奖励函数、OU噪声与变参数的负载,分级学习训练控制模块,迁移学习保留和巩固训练经验,仿真并验证使用深度确定性策略梯度算法的采煤机调高控制系统的效果。结果表明,系统输出曲线路径跟踪性极强,响应速度极快,输出更加平稳,震荡幅度小于0.01 m,与ACO-PID控制相比,震荡幅度减少50%,提高了控制系统抗干扰能力,有效地解决了采煤机滚筒调高控制系统响应速度慢、精度低、平稳性差的问题。Aiming at the problems of slow response speed of height adjustment,poor path tracking and vibration of drum and body,caused by large inertia,high load and complex and changeable working conditions during the cutting process of the shearer.A depth-based deterministic strategy based on gradient algorithm is proposed.Taking the cutting path as the environmental input of the height adjustment control system,which is obtained by fitting the sampling position of the demonstration knife,combined with the reward function,OU noise and variable parameter load,the training control module is learned hierarchically.The transfer learning retains and consolidates the training experience.The simulation and validation of the effectiveness of a shearer height adjustment control system are carried on by using a deep deterministic policy gradient algorithm.The results show that the system output curve path tracking is very strong,and the response speed is extremely fast.Compared with ACO-PID control,the output is more stable,the oscillation amplitude is 50%less,and the oscillation amplitude is less than 0.01 m,which improves the anti-interference ability of the control system.Altogether,using the strategy,the problems of slow response speed,low precision and poor stability of the shearer drum height adjustment control system can be solved effectively.

关 键 词：采煤机 滚筒调高 DDPG 分级学习 迁移学习 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TD632[自动化与计算机技术—控制科学与工程]