The performance of the ANNAI-based heading control system in heading and turning control is simulated on a mathematical ship model with the help of computer. The distance of the course from the ship position to the intended course is included in the learning process of the ANNAI controller. The performance of the ANNAI-based track control system is then demonstrated through computer simulations under the influence of external disturbances.
List of Tables
Nomenclatures
Introduction
- Background and Motivations
- The History of Automatic Ship Control
- The Intelligent Control Systems
- Objectives and Summaries
- Original Contributions and Major Achievements
- Thesis Organization
This adaptive neural network controller is then used to design ship tracking control system. Based on the proposed neural network control scheme, an automatic docking control system for ship is developed. Therefore, in this thesis an adaptive NN control system has been developed for ship control problems in direct methods and will be presented in chapter 2. The goal of this research is to develop an adaptive NNC for marine vessels.
Adaptive Neural Network by Adaptive Interaction
- Introduction
- Adaptive Neural Network by Adaptive Interaction
- Direct Neural Network Control Applications
- Description of the ANNAI Controller
The task of the NN is to "learn" the plant behavior from its current and previous states (through time delay operators z-m and z-n), and then to derive appropriate control actions in the next time step. In [72], the structure of the NN controller is derived using filtered error notations and passivity approximation. A uniform ultimate boundedness of the closed-loop system is given in the sense of Lyapunov.
- Training Method of the ANNAI Controller
- Intensive BP Training
- Moderate BP Training
- Training Method of ANNAI Controller
The flow diagram of intensive BP training method as in [89] and [90] is shown in fig. Next, the NN weights are updated using the BP training method so that the cost function Ek can be minimized. The flow chart of the training method of the proposed ANNAI controller is shown in Fig.
ANNAI-based Heading Control System
- Introduction
- Heading Control System
- Simulation Results
- With adaptation of n and γ
- Conclusion
In this subsection, a direct adaptive heading control system based on ANNAI is proposed for course keeping and yaw control. The task of the ANNAI-based steering control system is to find the appropriate steering wheel angle to minimize the following cost function. In addition, the algorithm for automatic adaptation of NN parameters is implemented in the ANNAI-based heading control system.
In the previous study we showed that the proposed ANNAI-based steering control system needs much less training iterations than the BPNN-based autopilot [57]. First in this subsection, an ANNAI-based steering control system is simulated in the case where the activation function of the output neuron is tangent sigmoid with fixed values of n and γ. Wind disturbances and measurement noise are used to test the performance of the ANNAI-based steering control system.
The ANNAI and BPNN based heading control systems have shown good performance with and without noise and disturbances. These simulations demonstrate the feasibility and effectiveness of the proposed ANNAI-based course control system. This comparison can be confirmed by numerical results of Eψ = 628, Eδ = 20 for ANNAI based course control system;.
Eψ = 660, Eδ = 33 for BPNN based steering control system. 3.11 shows the training process of ANNAI and BPNN within a control cycle.
ANNAI-based Track Control System
- Introduction
- Track Control System
- Simulation Results
- Modules for Guidance Using MATLAB
- M-Maps Toolbox for MATLAB
- Ship Model
- External Disturbances and Noise
- Simulation Results
- Conclusion
In this chapter, the improvement of the ANNAI-based runway control system is carried out by considering the off-track lateral distance from the target runway in the learning process of the ANNAI controller. The main results of this research were already published in [60], where the ANNAI-based track control system proposed in [58] was revised and the improved ANNAI-based track control system was presented. -Track distance is added to the NNC learning process to improve the adaptability of the ANNAI-based track control system to the effects of external disturbances.
In navigation practice, the precise track control performance of a ship is always affected by external disturbances such as ocean current and wind. To improve the track control capability, in this research, the distance of the track (see Fig. 4.2) is considered to design a SIMO track control system based on the ANNAI. Based on the above ANNAI controller and modified LOS algorithm, we propose a track control system as shown in Fig.
In this subchapter, simulation results of the ANNAI-based track control system are presented without and with the influence of measurement noise and external disturbances. 4.6, track control simulation of the ANNAI-based track control system without the influence of disturbances is shown. More simulation works and comparisons between conventional LOS-based and the proposed ANNAI-based track control system can be found in [60].
Our goal is to improve the ANNAI-based track control system, utilizing the learning capabilities of the ANNAI controller.
ANNAI-based Berthing Control System
- Introduction
- Berthing Control System
- Control of Ship Heading
- Control of Ship Speed
- Berthing Guidance Algorithm
- Simulation Results
- Simulation Setup
- Simulation Results and Discussion
- Conclusion
Using the advantages of NNC developed in our previous studies, this chapter presents an adaptive NNC and its application in automatic ship mooring control. The input signals of ANNAI1 are simply the direction error and its time-delayed values. Similar to ANNAI proposed in [60], the adaptation laws for hidden layer weights and output layer weights of ANNAI1 are respectively as follows.
I1 is the summation of the weighted inputs to the units in the hidden layer plus j. The configuration of ANNAI2 corresponds to ANNAI1 and is shown in Fig. The radius of the planned approach route is therefore not the same at every point on the route.
In practice, the ship's course is emphasized in the initial phase of the docking process. The effect of wind disturbance on the hull of the ship is based on the work of Isherwood (1972), introduced in [79]. The simulations show that the offshore and onshore winds influence the ship's lateral velocity and the final x-ordinate, but the robustness of the NNC is preserved.
The obtained simulation results lead to the following conclusions. a) KKN can be trained online without the need for any training data and the offline training phase.
ANNAI-based Dynamic Positioning System
- Introduction
- Dynamic Positioning System
- Station-keeping Control
- Low-speed Maneuvering Control
- Simulation Results
- Station-keeping
- Low-speed Maneuvering
- Conclusion
The propeller assignment block is used to calculate the contribution of each actuator of the ship's propulsion system. The adaptation law for the hidden layer of the ANNAI controller as in [58] can be written as By using the control scheme described in (6.9), the DP system can compensate for the unknown deviation representing slowly changing environmental disturbances and reduce the positioning error.
Further details of the ANNAI fitting laws were shown in chapter 2 and can be found in [58] and [59]. Suppose we want to make a certain point H(xH, yH) of the ship (as shown in Fig. 6.2) follow the desired path (stabilize at R). Here, the adaptation laws of ANNAI controllers are defined similarly as in the previous subsection.
In this simulation, the ship's center of gravity moves along the desired track a small distance off course and the course is maintained at the desired value. However, in this simulation, the bow of the ship can follow the desired track, while the ship's heading is kept at the desired value on each segment. To address this challenge, this chapter presented a novel hybrid neural adaptive DP system, which is independent of the exact mathematical ship model, using ANNAI controllers and a conventional PD controller.
Computer simulations are performed to prove the feasibility of the proposed controller and to test its performance.
Conclusions and Recommendations
- Conclusions
- ANNAI Controller
- Heading Control System
- Track Control System
- Berthing Control System
- Dynamic Positioning System
- Recommendations for Future Research
It also helped avoid the manual, time-consuming trial and error selection of neural network parameters. The proposed neural network takes less time to calculate the control output compared to the conventional backpropagation neural network. This ability is achieved by the online training program applied to the neural network.
Because the neural network controller can be directly adapted without approximating the ship dynamics, and no mathematical model of the ship was required in the controller design. To increase the adaptability of the track control system, the off-track distance from the ship to the target track was included in the learning process of the neural network controller developed in Chapter 2. Using the learning ability of the neural network, the system track control can adapt to changes in external disturbances as well as to ship dynamics. They can be used to calculate and display the ship's motion on the monitors of navigational equipment such as ECDIS.
The proposed neural network controller has been modified to be suitable for controlling the ship's heading and speed at low speeds in harbor maneuvering. The off-track distance from the ship to the predicted mooring path was included in the ship's heading control algorithm. This DP system is independent of the precise mathematical model of the ship and uses the proposed neural network controllers in parallel with the conventional PD controller.
For low-speed maneuvering, an algorithm was proposed to guide the ship along the intended track and maintain the direction of the ship.
Maneuvering and Control of Marine Craft, A volume of proceedings from the IFAC Conference, Brijuni, Croatia, September 10-12/1997, p. Theoretical Study). An adaptive autopilot for course keeping and tracking control of ships using adaptive neural network (Part II: Simulation Study).
Appendix A
Mathematical Model of Dynamic Positioning Ships
- Equations of Motion
- Bias Modeling
- Wave Force Modeling
- Measurement Systems
Unmodeled external forces and torques due to wind, currents and waves are lumped together into a ground-fixed constant (or slowly varying) bias term b∈ℜ3,. A common model for the prestressing forces in wave, pitch and yawing moment for marine vessel control applications is. This model can be used to describe slowly varying environmental forces and moments due to 2nd order wave loads, ocean currents, wind and unmodeled dynamics.
Wave forces can be divided into 1st-order wave disturbances and 2nd-order wave driving forces. For the practical application to control system design, the 1st-order wave disturbances can be described by three harmonic oscillators with some damping. For conventional ships, positions and yaw angles are usually measured by global positioning system (GPS) or hydroacoustic positioning reference (HPR) systems, and.
It is assumed that the total position of the ship can be obtained by superimposing the position and direction of the ship and the wave displacements.
Appendix B
Parameters used in the Simulations
Container Ship
A mathematical model for a single-screw high-speed container ship in surge, yaw, roll and yaw is shown in [78].
Multi-purpose Offshore Supply Ship
