This thesis is mainly composed of three parts, that are kinematic analysis, dynamic modeling and anti-sway/skew control of the container cranes. In the part of kinematic analysis, the closed-form position and velocity solutions of the spreader are derived by using the PSD(Position Sensitive Device) sensor that is useful to estimate the spreader motion in real time. The dynamic analysis is prepared for more exact modelling of the cable driven spreader including the stiffness of the cables. The derived dynamic model is approximated and used to model based compensation terms of the anti-sway/skew control of the spreader. In order to control the spreader motion, optimal tension resolution of the redundant cable driven mechanism should be needed. The simple graphical resolution method using null space of the Jacobian matrix is introduced for real time control, and it is applied to model based PID control of the spreader motion.
In general, the conventional container cranes operated by the skilled operator mainly adopt parallel 4-cables. Thus the sway and skew motion of the spreader certainly occurs in the travelling and traverse motion of the cranes. The operator has to adjust the spreader position by using joystick to prevent the misalignment between spreader and container or container and container or container and trailer.
The unmanned gantry cranes offer the solution to cope with the lack of skilled operator and the growth of container mobilization. However, the sway and skew motion of the spreader become serious problems in that case. To overcome those problems, the use of non-parallel auxiliary cables as an addition to the main cables is one of the proper methods.
The main objects of this thesis are focused on the research of the spreader motion control by using the auxiliary cables. To obtain the control methods and structures, the kinematics, dynamics and system modeling of the cable driven spreader system are analyzed in step by step.
The kinematic analysis of the multi-cable driven system has similar properties to well known parallel mechanisms. However, in the case of 6-legged parallel mechanisms, only two types of these have the closed-form solution. One is the type that the position and orientation in the kinematic equations can be decoupled and the other is that the base and moving platform have planar and similar figures each other. The kinematic solution of all of other types except for these two cases need to be obtained by the numerical iterative method.
The lengths of the 4-auxiliary cables are known but the spreader motion of 6-DOF can not be solved by using only these information. Thus, to obtain another two necessary information the PSD sensor is used. Therefore, the full 6-DOF kinematic equations of the spreader position and velocity are derived.
The dynamic equation of the spreader driven by 4-main cables that are assumed to have constant stiffness is obtained by using the Lagrange equation. The kinetic and potential energy are approximated through the decoupling of the translational and rotational motions of the spreader. The approximate dynamic equation has reasonable accuracy and it can be used as the plant model sufficiently to analysis the control algorithm.
The control to suppress the sway and skew motion of the spreader is performed by 4-auxiliary cables. The system become actuator redundant because the number of the auxiliary cables, actuator, is more than three control variables, x-y translation(sway) and rotation(skew). The redundancy is necessary condition to manipulate a thing by the unidirectional actuator such as a cable or magnetic field.
To resolve the tension of the auxiliary cables, i.e. optimal distribution of the tension to each cable, the graphical method of the null space of the Jacobian matrix is introduced. This method does not apply to more than two or three dimensions of the null space but is simple and capable for real time calculation that is difficult to obtain by other optimization method such as quadratic or linear programming.
The sway/skew control is performed by the model based PID control and its algorithm is composed of the conventional PID and the feed-forward terms. The feed-forward terms are derived by using the simplified and decoupled model of the spreader dynamic equation for next real application. The performances of the control including tension resolution algorithms are tested by the many kinds of computer simulation.
The remained topics of further research in the control of the container cranes are to study the efficient method to manage the tension limit of the cables and the fine motion control of the spreader.