Study on ship automatic berthing system with mooring lines

This study describes an automatic berthing system with mooring lines. It is designed to be berthed by using mooring device on the upper deck of a ship. It is to berth once maintaining parallel with the quay by controlling both forward and aft breast lines. Berthing method is used through length adjustment of mooring lines connected between ship and quay by controlling the angular velocity and the torque of hydraulic motor in mooring device. The study is conducted under three changing conditions of draft, such as even-keel, rise of the gravity center and trim to stern. Variables affecting berthing stability are determined based on the control performance of each condition. Bond graphs method is used to model the system. Controller is designed as PID control method of reference-model algorithm. The control program is composed of synchronous control system based on the equations derived with the numerical analysis. The tank test is conducted to verify the usefulness of the control program.


Introduction
The system is designed to realize automatic berthing by adjusting mooring lines using hydraulic winch equipped on the upper deck of a ship. In general, there are two methods to berth the ship to the quay. One is to move obliquely at low-speed toward the quay using autopilot with sidethrusters at a certain point, and the other is the traditional method to move laterally by using tugboat and mooring lines. The mooring line in these methods is rarely used for berthing, and it is used only to fix or limit the position of ship within the required range. Most ships use the traditional berthing method with mooring line and tugboat, even though ship control techniques have been improved. These methods have trouble to ensure the berthing stability, and there is no other option but to increase the number of tugboat in proportion to the port's characteristics and the ship's size. It increases the berthing costs in the end. In addition, the frequent berthing from short voyage needs skilled sailors to maintain berthing stability, but it is getting harder to find skilled sailors. For these reasons, the berthing system can deal with risks as errors of judgment with the absence of worker while considering economic aspects and ease of application.
Berthing assistant system has been researched using the concept of dynamic mooring system since side thruster and autopilot were applied in the berthing process. The dynamic mooring system is, as control system for path tracking, the control method for a precise follow-up of the path by setting movement path of the target point in the process of mooring or berthing. The related control technologies are the self-turning adaptive control (Kalman filtering), LQG optimal control, H∞ control, fuzzy control, artificial neural networks, and fuzzy-sliding mode control and so forth, and the performance continues to be improved steadily (Mort and Linkens, 1980;Zuidweg, 1981). Among them, an optical trajectory planning control method is one of the most appropriate technologies to cope with various berthing environments. It has controlled rudder angle and propeller by programming the best path of berthing and has enabled maneuverability and obstacle avoidance control by keeping the path between certain points and berthing point through off-line path planning. It has also improved the performance by applying ANN control and re-analysis in parallel with the operation of a tugboat, and a method to use GPS (Djouani and Hamam, 1994;Hasegawa and Fukutomi, 1994;Ueno and Santerre, 2000). However, this method needs autopilot and side thruster to ensure the mooring stability. Most existing ships still stick to the traditional meth-od using mooring line and tugboat by berthing. The latest control technique should be applied to improve the berthing method but it is difficult because the installation of necessary devices to precise control increases time costs. Even if the devices have already been installed, the operation efficiency is not uniform according to the operation ability of the workers and port environment. In order to overcome these problems, we propose an effective berthing system which uses mooring devices of a ship. It is the ship's automatic berthing or approaching system (SABS) by mooring lines. This system has not been proposed so far by other researchers. It is easy to apply to all ships and ensures a uniform berthing capability and can improve the efficiency and stability of the berthing process.

Research contents
In order to design the berthing system, the dynamic characteristics of a ship should first be analyzed. It is determined by six degrees of freedom composed of rotational and translational motion depending on the berthing purpose and the external environments. Many studies have been analyzed based on the degree of freedom in the movement by considering only interference phenomenon between the external environment influence and the ship movement (Aryanpour and Ghorashi, 2001;Falzarano et al., 2001;Kong et al., 2004;Natarajan and Ganapathy, 1995;Wang et al., 2011). It aims to analyze the stopped ship by considering the interference phenomenon between the external environments and the ship movement, and it shows good results (Chang et al., 2012;Das and Das, 2006;Wang et al., 2000).
The proposed system is to berth once the ship maintains parallel to the quay using mooring lines from the specific point. The study focuses on the systematic viewpoint of the berthing device and the precedent study for completion of the system. The ship generally has six mooring lines as shown in Fig. 1, but this study is conducted to use two breast lines of the forward and aft. In general, the berthing port is of relatively less influence on the ship from the external environment, and the berthing velocity is low, speed of about 0.05 m/s on the average and smaller than 0.1 m/s on the maximum (Association, 2005). There is little heave motion because the ship is towed in parallel with the quay based on the pulling rate of the breast lines by both hydraulic motors. The heave motion can be neglected for these reasons, but it is still considered because it is required to calculate the rolling motion. The effects of wind and tidal current are ignored. Surging motion is ignored because it is offset by ahead or astern operation of the propeller and has no consideration for stern line with bow line to control it. Therefore, the ship dynamics are covered in the effects of two breast lines of forward and aft, and it is analyzed based on four degrees of freedom, such as swaying, rolling, heaving and yawing.
Experimental preconditions of the study are as follows: -The supply pressure for the hydraulic system is constant without its internal leakage.
-The resistance generated in hydraulic lines is ignored.
-A wetted surface area is constant.
-The direction, tension, and thickness of the mooring line do not affect the ship.
-The effect of waves acting on the ship is ignored. We perform simulations and experiments under three ship conditions.
-Condition I is an even-keel state.
-Condition II is a rising state of the gravity center from an even-keel state.
-Condition III is a trim state to the stern. This study is summarized in four procedures. First, it analyzes the mutual influences between the nonlinearity of the fluid and the ship dynamics. Second, it designs the model for the automatic berthing with the dynamical identification of a ship and develops an optimal control algorithm based on the designed model. Third, it designs a controller and compares the design values with the results of the simulation. Finally, it performs berthing control for parallel movement of a ship and verifies the effectiveness of the system through a tank test. In the numerical analysis, it derives quantitative values by considering the friction and viscosity of the fluid affecting the movement of the ship and estimates the variables that affect ship's dynamical characteristics depending on the operation of the hydraulic system. The model is designed to track the position of a ship over time using bond graphs modeling method. The control algorithm is applied to a practical reference-model algorithm that the present authors have developed (Hur and Yang, 2007), and the control program is designed with a synchronous control. We conduct a tank test for control experiments and analyze the experimental results through comparison with the simulation, and then measure the parameters that affect the dynamic characteristics of a ship.

Numerical analysis and modeling
3.1 Hydraulic and mechanical system SABS is composed of hydraulic, mechanical, and dynamical systems. The hydraulic system consists of hydraulic pumps, valves, and hydraulic motors, and the mechanical system consists of reduction gear and drum. High-speed on/off solenoid valve is used to control, and pulse width modu- (1), which shows the percentage modulation (τ [%]). The variable t on is the starting time of the input pulse, and t max is the modulation period. Eq.
(2) represents the flow rate through the load port of the valve (Q valve ): where, A valve is the opening area of the valve, C d is the flow coefficient of the valve, ρ oil is the oil density energy of a system, P s is the pump pressure, and P m is the motor pressure. Based on Eqs.
(1) and (2), the changes in the micro-flow rate of the percentage modulation can be expressed by Eq.
where, K valve is the gain of the valve. The gain K valve must be determined in the basic experiments because the flow rate through the valve according to the modulation speed depends on the supply pressure. It is the gain that determines the berthing velocity, and it is needed to find the value close to the average berthing velocity through repeated experiments. The percentage modulation is calculated in real-time by the controller in the control experiment, and it is converted to the digital signal.
The hydraulic system of the mooring device is composed of the hydraulic pump, control valve, hydraulic motor and the reduction gear to transmit the power. The power is converted into the angular velocity and torque through the reduction gear, and then is adjusted to the moderating ratio of the gear. The angular velocity (N gear ) and the torque (T motor ) generated by the hydraulic pump are determined by considering the moderating ratio of the reduction gear, and they are obtained from Eq. (4). T drum = N gear ¢ T mot or ; drum = mot or =N gear ; (4) where, T drum is the torque of the drum connected to the hydraulic motor, N gear is the torque of the reduction gear, θ drum is the angular velocity of the drum connected to the hydraulic motor, and θ motor is the angular velocity of the hydraulic motor.

Ship dynamic system
The ship dynamics are analyzed in terms of three degrees of freedom (swaying, rolling and yawing) in consideration of the Y-axis direction component because this system is designed to move the ship in the lateral direction.
The surface frictional force of the ship and the fluid force affect the hull at a constant speed for the draft states of the ship. It works in the tangential direction to the hull by viscosity of water, and the fluid force is changed by the wetted surface area of the hull. The external forces as airflow also add to it depending on the topside shape of the ship. In general, equation of the surface frictional force of the ship is calculated by Eq. (5) in non-dimensional friction resistance relations.
where, R f is the surface frictional force of the ship, C f is the friction coefficient for the surface, ρ is the density of the fluid, S s is the wetted surface area, and v is the velocity of the fluid. The friction coefficient for the surface (C f ) is given by the ITTC 1957 model-ship correlation curve formula, and the wetted surface area (S) is defined by Olsen's equation. And they are expressed in Eqs. (6) and (7).
velocity, ν is the coefficient of kinematic viscosity, and L s is the ship length.
where, C s is the fitness coefficient, B s is the extreme breadth, D s is the draft, and C b is the block coefficient. The resistance force of the wind is expressed by the direction that affects the hull, such as the bow, the stern and the side. The air resistance R air is in Eq. (8), and the method proposed by Isherwood (1972).
where, C air is the air resistance coefficient, θ is the angular displacement, and D 1s is the freeboard. The inertia moment due to the lateral movement I s is expressed in Eq. (9), and the ship shape is considered as cubic.
where, M s is the ship mass. This system needs calculations of the rolling degree (ψ) to ensure the berthing stability from the influence of the mooring line. Heaving can be obtained by the relation between the ship weight and the pulling force of mooring lines. The rolling degree here can be expressed as a function of the change in velocity by the relationships of the lateral (Y) and vertical (Z) directions of the hull, and it can be expressed in Eq. (10). The inertia force of the ship must be considered an upper limit value of the equivalent velocity of the ship to ensure the berthing stability in Eq. (10). This can be obtained in Eq. (11), and it determines a berthing velocity to minimize the inertia force of the ship.
In order to determine the berthing distance and the ship attitude, Eq. (12) can be obtained by the correlation between the overall length and the distance between the quay and both sides of the ship.
where, S y is the quay-center distance, S y1 is the quay-forward distance, S y2 is the quay-stern distance, L 1 is the center-forward distance, and L 2 is the center-stern distance.
The yawing degree (ϑ) exists when berthing can be obtained in Eq. (13) by utilizing the difference in the distance between both sides of the ship and the quay.
3.3 System modeling Bond graphs method is useful for the system modeling because it can obtain linear equation by linearizing the nonlinear system, and can simply add or remove a specific system combined with the mechanical, hydraulic, pneumatic systems, etc. (Karnopp et al., 1997).
In Fig. 2, the right part represents the forward side, and the left part represents the aft side. The junction and the parameter of each node show the characteristics of the system. Based on Nodes 9 and 29, the upper part represents the hydraulic system among the mooring devices, and the lower part represents the dynamic characteristics of the ship by the influence of the mooring line. In the upper part, R of Nodes 3 and 23 represents the dynamic characteristics of the valve in the hydraulic system, and it can be obtained through Eqs.
(2) and (3). The detailed analysis of the characteristics has been described in the previous study of the author (Yang et al., 1999;. The dynamic effect of the control valve is very slight on the numerical analysis because the operation velocity of the system is very slow. R of Nodes 8 and 28 represent the loss of pipe friction, and R of Node 8 represents the flow returning to the tank. The transformer TF represents the dynamic transformation in the interaction between the systems. Transformer TF of Nodes 9 and 10 or Nodes 29 and 30 signifies the hydraulic motor, and represents the dynamic relations of the flow rate, which is converted to the angular velocity of the hydraulic motor through the valve and the mooring lines. Transformer TF of Nodes 11 and 12 or Nodes 31 and 32 represents the reduction gear, and transformer TF of Nodes 13 and 14 or Nodes 33 and 34 represents the drum. The transformer variables M, m, D, d, G, and g are the physical property values for combining each system, and the torque and angular velocity in Eq. (4) can be obtained by using these values. M and m represent the dynamics between the valve and the motor, and G and g represent the dynamics between the hydraulic motor and the reduction gear. D and d represent the dynamics between the reduction gear and the drum. The lower part shows the dynamic system of the ship. R of Nodes 15 and 35 represent the frictional resistance force acting on the hull, and the characteristic equations are given by Eqs. (5) Transformers L and l represent the dynamic characteristics between the drum and the change in the length of the mooring line, and transformer S represents the dynamic characteristics between the change in the length of the mooring line and the motion characteristics of the ship. The result of the modeling is derived from I based on the continuous calculation results of each characteristic equation. In junctions 1 and 0, it is to obtain the junction equation, and it helps to know the influence degree of each parameter in the whole system.

Basic experiment
This system was modeled on a chemical tanker to sail along the coast. Table 1 shows the principal particulars of  the experimental object. The model ship is of a scale of 1/100 to the actual ship, the valve is a 2-port 3-way valve (HS-G01-A21), the hydraulic motor is an orbital motor (OMM 20) and the distance sensor is a non-contact ultrasonic sensor (TS-30S). The pressure sensor (PMS, Sensys) is equipped in the control port side of the valve. Fig. 3 shows the structure of the experimental apparatus. The basic experiments were conducted by setting the same system as the mooring device of the ship on the outside of the tank. The distance sensors built in the forward and aft measure the distance between the quay and the ship, and the measured signals are analyzed by the control program through an A/D-D/A board and amplifier. The analyzed results control the opening of the valve of both sides through the feedback loop, which is retransmitted to the switching board and A/D-D/A board. At this time, the pressure difference due to the flow rate changes is detected, and the hydraulic motor is operated simultaneously. The rotation radius of the drum connected with the hydraulic motor is controlled depending on the sensing distance between the quay and the ship. Therefore, the synchronization controller controls the distance from both sides of the ship to the quay continuously, and it keeps the ship parallel to the quay. Fig. 4 shows the results of the basic experiment that has been done to estimate the gain in determining the berthing velocity. In general, the average berthing velocity is about 0.05 m/s, and it decreases with the increasing tonnage (Association, 2005). We conducted the basic experiment to estimate the gain of the reference-input, which corresponds to the average berthing velocity. It was done under the condition in which the ship was under Condition I and the distance to the quay was a constant. The ship was in a stop state in the fixed distance, and the hydraulic motor had a constant velocity maintained. 1 V means 0.1 m. The gain corresponding to the average berthing velocity is 0.6. It determines the berthing velocity by limiting the opening of the valve in PID controller and it means K valve of Eq. (3).

Controller design
The control program was designed based on the results of the numerical analysis. The controller was designed to a synchronous control system using the practical referencemodel algorithm developed by Hur and Yang (2007). Fig. 5 shows the control diagram of SABS. It is made of the synchronous control system to control the forward and aft side of the ship. The structure is intended to transmit the data in the computer through AD-DA card after the distance sensors measuring the distance between the quay and both sides of the ship. The measured data are used to calculate the berthing distance to the quay, and a control input signal follows the reference input on the control algorithm, and then the control output controls the ship attitude and the berthing velocity. Fig. 6 shows the structure of the controller (Hur and Yang, 2007).
In order to obtain the reference-model, the pre-filter should be obtained with the model of the control object. It is described in detail in the reference literature, and the results    K. U. Yang et al. China Ocean Eng., 2017, Vol. 31, No. 1, P. 19-29 can be obtained as Eqs. (14)-(17). Eq. (14) is the obtained model to bond graphs modeling, and it is the third-order control system G p (s).
G m (s) = 1 0:0111s 4 + 0:065s 3 + 0:310s 2 + 0:647s + 1 : (15) Pre-filters G f (s) and G c (s) can be obtained as shown in Eqs. (16) and (17) Based on the calculated results, Fig. 7 shows the control model with the reference model. The control model means the control response on the control program and it should follow the reference model. The reference model was designed to reach the reference input as quickly as possible by minimizing an initial dead band time of the control object and considering the inertia force of the ship.

Simulation
The simulation was done to estimate the dynamic characteristics about each variable that affects the system. It was done with the gain obtained from the basic experiment. The period and length of PWM and the input signal were set regularly. Table 2 shows the parameters and values used in the numerical analysis, the simulation and the control experiment. Fig. 8 shows the responses of the mooring line force with the position under Conditions I(A) and II(B). The position means the straight distance between the quay and both sides of the ship. Y-axis of each response is the negative sign because the initial value was set on the berthing point basis. Each response shows the nonlinearity. As looking at the portion about 0-13 s in the responses, the distance is stagnated, and the mooring line forces indicate that the slope inversed to the positive direction from the negative direction. It is because the ship has the vertical force caused by self-weight early while it begins the berthing. It means that the static friction force of the ship exists at the early moment of the berthing. In other words, the time zones show conflicting states between the static friction force and the mooring line force. It makes the occurrence time of the inertia force with the dissipation time of the static friction force to be estimated. A Comparison of the results under Conditions I and II shows that under Condition I the delayed response of the simulation full-time is about 10 s, and the inverse time of the mooring line force is about 15 s, with a delay of 2 s. It is caused by the increase of the gravity center in the ship. The increase of the ship weight increases the fluid resistance with the increase of the wetted surface area, and it means that the static friction force affects the delay time. It also can be seen that the delay time in the early stages affects the whole response velocity of the berthing. Fig. 9 shows the responses of the distance and the mooring line force under Condition III. Fig. 10 compares the responses of the distance in Fig. 9. Fig. 11 shows the responses of the roll degree, the yaw degree and the inertia    Fig. 9, the forward's is about 17 s, the midship's is about 12 s and the aft's is about 11 s. In Fig. 10, the gap of responses goes wider as it goes toward the aft. It can be known that the gaps are closely connected with the inversion time in Fig. 9. The gravity center moves toward the aft side under Condition III. It increases the wetted surface area of the forward, and increases the inertia moment of the ship by causing unbalance of the ship weight between the forward and the aft. As a result, the fastest response shows in the forward side. In addition, the midship and the aft show a tiny gap compared with the forward because the gravity center moves to the aft according to the longitudinal metacentric height. It can be known that the longitudinal metacentric height determines the fluid resistance if the trim is changed. Therefore, these results show the effect of the wetted surface area with the inertia moment if the ship is under Condition III showing the movement of the gravity center. It also tends to increase overall the yaw and roll of the hull by the inertia moment when the ship comes alongside the quay under Condition III. The mooring line force needs to be adjusted separately both mooring lines to offset the tendency. Fig. 12 shows the response comparison of the rolling degree under each condition. The rolling degree shows the smallest response under condition II. It is found that the stable berthing is possible if the gravity center is increased uniformly in each side. In the comparison of the results under Conditions I and III, the rolling degree of Condition I must be larger than that of Condition III according to the size of the wetted surface area. However, the gravity center is located at the aft under Condition III, and due to its location the moment of inertia amplifies with the aft as the center. That is, the enlarged chart of Fig. 12 shows the results that the rolling degree of Condition III is larger than that of Condition I. Figs. 13 and 14 show the response comparison of the position under all conditions. The berthing velocity of both     K. U. Yang et al. China Ocean Eng., 2017, Vol. 31, No. 1, P. 19-29 25 sides is increased in proportion to the difference of the wetted surface area. Condition II shows the slowest velocity in the responses. From the enlarged graphs of Conditions I and III, Condition III shows the fast response velocity in the forward, but Condition I shows an opposite result in the aft. Each gap shows the increased response in the aft particularly. It is because the fluid resistance force works in proportion to the wetted surface area of both sides. It can be known that the difference of the wetted surface area in both sides affects the ship attitude during the berthing. Therefore, the mooring line force and the berthing velocity should be adjusted by considering the wetted surface area when the ship is under the trim condition to stern or under the condition of random trim.
In the results of the simulation, it could be known that the mooring line force and the berthing velocity need to be adjusted appropriately to ensure the berthing stability. The parameters affecting the dynamic characteristics of the ship were distinguished depending on each condition of the ship. Condition II showed the most stable response in view of the rolling degree and the berthing velocity. Under Condition III of the most unstable state, the yawing was the cause of amplifying the rolling of the ship, and it occurred by the difference of the wetted surface area in the forward and aft of the ship. To determine the whole berthing time was the static frictional force caused by the self-weight of the ship at the early stage of berthing. Fig. 15 shows the experimental device of SABS. It was composed of the hydraulic system, the sensor, the computer and the circuit board for signal processing. The experiment was conducted with installing the same system on the outside of tank since it is difficult to install the experimental devices on the model ship. The early attitude of the ship is on an arbitrary state because the ship is difficult to maintain parallel to the quay by the arbitrary inertia of the ship or the slackness of mooring lines. The control conditions of the system are that the ship should be berthed while keeping parallel to the quay, and the degree of the ship attitude should be within the range of ±2° in the parallel state to the quay. The outputs are expressed as voltage, 1 V of 0.1 m. The range of the control output signal is 0-0.5 V. If the ship attitude is out of the set range, it is adjusted by the control program. The distance between the ship and the quay is 0.6 m, the experiment time is 80 s, and the sampling time is 0.01 s. The overshoot appearing on the responses was generated by a disturbance signal of the distance sensor by the high frequency, and it was caused by the high frequency signals from the frequency converter in the hydraulic system. The control program was designed to handle the overshoot since this phenomenon may occur in the actual berthing environments. Fig. 16 shows the control responses of the distance between the ship and the quay under Condition I. The control point is PWM, and it indicates the point and intensity of the controlled variable to control the berthing and the ship attitude. Up to about 20 s, the responses are a positional deviation between the forward and the aft, but they represent a   steady state to maintain parallel over time. The static friction force mentioned in the simulation is difficult to distinguish, but it can be estimated by comparing the slope of the early stage with the total response. It shows the unbalanced position between both sides of the ship and the quay when the mooring line force has influence on the ship at the early stage of berthing. This is because it is difficult to set the same length because of slackness of the mooring lines and hard to maintain parallel between the ship and the quay at the early stage of berthing. Thus, the control program was designed to be concentrated more on the attitude maintaining control than the berthing control at the early stage. As a result, the responses have shown the gentle slope until about 20 s, and they drew the good results following the designed value of the berthing velocity to maintain parallel to the ship. It also showed the output range of 0-0.5 V based on the setting condition of the attitude. When the berthing is started by the mooring lines, the inertia of ship is generated by the mooring line force, and then the ship moves by itself. However, the intensity of inertia is dependent on the mooring line forces. The berthing velocity cannot be adjusted if the intensity of the forces is too strong. The ship is at the risk of collision in the quay because the berthing velocity increases by the inertia. Therefore, it is necessary to adjust precisely the radius of the drum to prevent being uncontrolled by the inertia. That is, the opening of the hydraulic valve should be precisely adjusted in consideration of the attitude and the berthing velocity during a short time around the occurrence of the inertia. Overall, the responses show linearity with the stable berthing velocity unlike the results of the simulation. Fig. 17 shows the control input by PWM in each mooring line of Fig. 16. The Y-axis is displayed as the opposite sign to compare both sides. The control input means the controlled variable of each mooring line, and it determines the timing and intensity of PWM. The controlled variable is concentrated on the early stage of each response. It shows that the control program is designed to implement the selective control of the berthing and the attitude, because the length of mooring lines is not constant by slack, and the initial attitude of the ship is in the unbalanced state. Because of this, even though the distance between both sides and the quay was measured equally, the controlled variables cannot be changed. Fig. 18 shows the responses under Condition II, and Fig.  19 shows the control input by PWM on both sides of Fig.  18. The responses show the similar result under Condition I of Fig. 16, and the berthing velocity just decreases in proportion to the increase of the gravity center. Fig. 20 shows the responses under Conditions I and II. The significant difference in the responses can be found in the occurrence time of inertia. For Conditions I and II, they are about 15 s and 20 s, respectively. It shows that Condition II affects the berthing velocity during the occurrence time of inertia. The slope of the responses maintains constant after the inertia occurs. That is, the increase of the Fig. 17. Control input by PWM on both sides of Fig. 16.   K. U. Yang et al. China Ocean Eng., 2017, Vol. 31, No. 1, P. 19-29 gravity center increases the fluid resistance force and the wetted surface area, but this result does not affect the berthing velocity after the inertia occurs. Fig. 21 shows the responses under Condition III. Fig. 22 shows the control input by PWM on both sides of Fig. 21. The velocity on both sides by the intensity of mooring lines is the same at the early stage of berthing. However, Condition III increases the wetted surface area in the midship and the aft if the gravity center is moved towards the aft. It also increases the fluid resistance force toward the aft. At this time, the inertia moment rotating around the aft occurs, and the forward side tries to move in the opposite direction to the quay. For this reason, the controlled variable of mooring line force in both sides should be controlled as a different intensity under Condition III. It can be known by the response that the forward is measured closer than the aft in the time of about 0-17 s in Fig. 21. It is because the control of the ship attitude maintenance is done on the forward side while the berthing control is done on the aft side in accordance with the change of the wetted surface area. This can also be confirmed by the control input of PWM in the same time zone of Fig. 22. In Fig. 21, the actual berthing starts with the occurrence of the inertia at about 17 s, and the berthing velocity increases gradually. The berthing velocity of the aft seems faster than that of the forward until about 55 s, and then the position of both sides is maintained uniformly. This shows that the control of both sides is applied selectively. That is, the berthing control is done on the aft side when the control of the attitude maintenance is done on the forward. Fig. 23 compares the responses under Conditions I and III on each side. In the responses of the forward and aft, the aft has been delayed until about 17 s, and the forward is delayed in about 17-50 s. It shows the results from the difference of the wetted surface area on both sides. With the occurrence time of inertia about 17 s, the aft shows the delayed response because the wetted surface area is increased under Condition III. However, the responses of both sides are not significant different. The forward of Condition III shows the delayed response with the occurrence time of inertia about 50 s. This is because the forward tends to move in the opposite direction to the quay by the inertia moment, and the berthing control is done in the aft.

Control experiment
The control program showed a good performance in response to each condition of the experiment. The performance of the system was dependent on the occurrence time of the inertia, and the berthing stability was dependent on the control method of both sides at the early stage for berthing. Therefore, the system must adjust the berthing velocity with the intensity of the mooring line force with considering the inertia occurrence time, static friction force, ship attitude and draft.

Conclusions
The system was devised to berth automatically with mooring lines using the mooring winch equipped on the upper deck of ships. We considered that the system is useful as the berthing system. It is possible to obtain cost-saving effects of the additional equipment with the mooring device equipped on the upper deck of ships and can be applied easily to the traditional berthing system. It can be used for the assistant device to keep a constant position. It can also be   K. U. Yang et al. China Ocean Eng., 2017, Vol. 31, No. 1, P. 19-29 used for the auxiliary device in the automation ship which has the autopilot and side-thruster. The rolling and yawing can affect the berthing stability. The rolling affected the berthing velocity under Condition II of the ship. The yawing affected the ship attitude under Condition III. The static frictional force occurred at the early stage for berthing, and the berthing velocity showed the difference depending on the draft. However, once the inertia occurred and the fluid resistance force was canceled out, the berthing was done linearly. The most important control element to be considered in this system was the trim by the stern and the inertia of the ship.
The next study will be analyzed based on the angle changes of lines connected to the quay including all lines with head line, stern line and spring line. It will also be analyzed in terms of six degrees of freedom by considering the environmental elements.