Fig 1.
Structure diagram of the self-adaptive height adjustment hydraulic system of the shearer.
1-Height adjusting hydraulic cylinder; 2-balance valve; 3-electro-hydraulic proportional directional valve; 4-height adjusting oil pump; 5-overflow valve; 6-filter; 7-oil tank; 8-detecting device; 9-rocker arm.
Fig 2.
The block diagram of the height adjustment control system for the shearer.
Fig 3.
Transfer function block diagram of the self-adaptive height adjustment control system.
Table 1.
Simulation parameters.
Fig 4.
Simplified schematic diagram of shearer height adjustment hydraulic system.
Fig 5.
Technical route of the adaptive height adjustment control process for the shearer.
Table 2.
The main structural parameter values of the shearer and drum.
Table 3.
Typical working conditions.
Fig 6.
3D solid model of shearer height adjustment mechanism.
Fig 7.
Coal wall model of working condition 4.
Table 4.
The signal characteristic value of vibration of drum X, Y and Z.
Fig 8.
Vibration acceleration curves of the spiral drum under cutting conditions of f = 3.5 coal and f = 3.5 Roof +coal.
Fig 9.
Basic principle of SVD denoising.
Table 5.
Parameter setting of CWT.
Fig 10.
Generation of frequency spectrum of drum vibration acceleration.
Fig 11.
Time-frequency spectrogram under the f = 3.5 pure coal condition.
Fig 12.
Time-frequency spectrogram under the f = 3.5 roof+coal condition.
Table 6.
Main parameter assignment.
Fig 13.
AlexNet transfer learning model.
Table 7.
Factor level table.
Table 8.
Test configuration scheme and orthogonal test results.
Table 9.
Table of factor influence degree analysis.
Fig 14.
Factor trend analysis.
Table 10.
AlexNet network migration learning model recognition accuracy.
Table 11.
Recognition accuracy and recognition time under different models.
Fig 15.
Algorithm architecture of the hydraulic height adjustment system for the shearer based on DDPG.
Fig 16.
Simulink model of height adjustment system.
Fig 17.
Deep neural network-Critic network.
Fig 18.
Deep neural network-Actor network.
Table 12.
Parameter setting of deep neural network.
Table 13.
Agent parameter settings.
Table 14.
Characteristic parameters of different reward functions.
Fig 19.
The DDPG-based self-adaptive hydraulic height adjustment system model for the shearer (Model I).
Fig 20.
Shows the system’s control performance under reward function r1.
Fig 21.
Shows the system’s control performance under reward function r2.
Fig 22.
Shows the system’s control performance under reward function r3.
Fig 23.
Tracking simulation of harmonic signal.
Fig 24.
Tracking simulation of square wave signal.
Fig 25.
Simulation analysis under disturbance condition.
Fig 26.
Environmental self-adaptability simulation analysis.
Fig 27.
Comparison of control effects between DDPG and classical controllers.
Table 15.
Comparison of control performance of DDPG and other algorithms.
Fig 28.
The AMEsim model of the hydraulic system of the shearer electro-hydraulic proportional height adjustment.
Fig 29.
Self-adaptive hydraulic height adjustment model of shearer based on DDPG (Model II).
Fig 30.
Piston displacement tracking and error.
Fig 31.
Piston motion speed.
Table 16.
Comparison of control performance between DDPG and typical deep reinforcement learning algorithms in the joint simulation environment.
Fig 32.
Technical route as determined by similar parameters.
Fig 33.
Self-adaptive height adjustment test system platform.
Fig 34.
Simulated coal wall model.
Table 17.
Similarity coefficient of test bench and coal wall.
Table 18.
Simulink mean error of model simulation and test platform test results.
Fig 35.
Hydraulic cylinder piston displacement test results and back deduction results data.
Table 19.
Verification of the stability and reliability of experimental results for the AlexNet transfer learning model.