목차

1. Introduction 1
1.1 Background on the generation of the optimal route 1
1.2 Background on route following control 3
1.3 Composition of the thesis 7
2. Dynamic Vessel Model 10
2.1 Coordinate systems 10
2.1.1 Definition of earth-fixed coordinate system and body-fixed coordinate system 10
2.1.2 Transformation between body-fixed coordinate system and earth-fixed coordinate system 11
2.2 6 DOF nonlinear vessel equation of motion 13
2.2.1 Rigid-body vessel equation of motion 14
2.2.2 Hydrodynamic vessel equation of motion 16
2.3 3 DOF nonlinear vessel equation of motion 23
2.3.1 Forward speed model 24
2.3.2 Maneuvering model 25
2.4 Linear 3 DOF vessel equation of motion 26
2.4.1 Linear forward speed model proposed by Blanke 27
2.4.2 Linear maneuvering model proposed by Davidson and Schiff 29
2.4.3 Combination of forward speed model and maneuvering model 30
2.5 Specifications of a vessel used in this study 32
3. Generation of the Optimal Route Based on Reinforcement Learning 33
3.1 Background 33
3.2 Related work 36
3.2.1 Concept of the reinforcement learning 36
3.2.2 Types of reinforcement learning algorithms 37
3.2.3 Selection of reinforcement learning algorithm 39
3.3 Considerations for generating the optimal route 46
3.3.1 Safety considerations for optimal route generation 48
3.3.2 Minimum fuel consumption considerations for optimal route generation 50
3.4 Optimal route generation using Q-learning (Busan port to Gamcheon port) 53
3.4.1 Environment setting for optimal route generation 53
3.4.2 Simulation conditions 59
3.4.3 Simulation results 61
3.5 Optimal route generation using Q-learning (Busan port to Busan new port) 64
3.5.1 Environment setting for optimal route generation 64
3.5.2 Simulation conditions 66
3.5.3 Simulation results 67
4. Velocity Type Fuzzy PID Autopilot Controller 69
4.1 Background 69
4.2 Limitations of the PD autopilot controller 71
4.2.1 Discretized vessel model used for the simulation 71
4.2.2 Simulation condition 72
4.2.3 Simulation results for the route following 73
4.3 Design of the velocity type fuzzy PID autopilot controller 77
4.3.1 Fuzzification algorithm 78
4.3.2 Fuzzy control rule 80
4.3.3 Defuzzification algorithm 82
4.3.4 Simplified control rule of fuzzy PID controller 84
4.3.5 Stability analysis 85
4.4 Performance verification of velocity type fuzzy PID autopilot controller 86
4.4.1 Simulation results for route following 86
4.4.2 Accuracy analysis for route following 90
5. Generation of the Environmental Disturbances 93
5.1 Background 93
5.2 Winds 94
5.2.1 Generation model for winds 94
5.2.2 Wind force and moment 95
5.3 Waves 100
5.3.1 Generation model for waves 100
5.3.2 Frequency of encounter 102
5.4 Ocean currents 104
5.4.1 Generation model for ocean currents 104
5.4.2 Ocean current applied to the vessel 105
5.5 Vessel equation of motion with environmental disturbances 107
5.6 Route following of the vessel with environmental disturbances 108
5.6.1 Discretized vessel model used for route following 108
5.6.2 First case 109
5.6.3 Second case 113
5.6.4 Accuracy analysis for route following 118
6. State Estimation of Vessel Based on the Kalman Filter 120
6.1 Background 120
6.2 Stochastic vessel model including white Gaussian noise 121
6.3 Fuzzy PID autopilot controller based on the Kalman filter 126
6.3.1 Kalman filter algorithm 126
6.3.2 Velocity type fuzzy PID autopilot controller using the separation principle 129
6.3.3 Simulation results applying the separation principle 131
6.4 Innovation process characteristics of the Kalman filter 135
6.4.1 Stochastic state space model of vessel including environmental disturbances 135
6.4.2 Innovation process when environmental disturbances are applied to the vessel 136
7. Estimation of Environmental Disturbances using a Fuzzy Disturbance Estimator 139
7.1 Determining the presence of environmental disturbances 139
7.2 Fuzzy disturbance estimator 141
7.2.1 Fuzzification algorithm 142
7.2.2 Fuzzy estimation rules 145
7.2.3 Defuzzification algorithm 146
7.2.4 State estimation algorithm based on Kalman filter combined with the fuzzy disturbance estimator 149
7.3 Velocity type fuzzy PID autopilot controller based on the fuzzy disturbance estimator and Kalman filter 152
7.3.1 Autopilot controller of the vessel with fuzzy disturbance estimator 152
7.3.2 Converting estimated environmental disturbances into the thrust and rudder angle 153
7.4 Simulation of environmental disturbance estimation and route following 156
7.4.1 Simulation of the first case 156
7.4.2 Simulation of the second case 163
7.4.3 Accuracy analysis for route following 169
8. Conclusion 171
References 174