Combining deep deterministic strategy gradient algorithm and hierarchical learning for coal mining machine height adjustment control strategy
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.
coal mining machineadjusting the height of the drumDDPGgraded learningtransfer learning
NETL
NSTL
万方数据
