As noted above, it is shown that effective dynamic positioning for the FPSO and the shuttle tanker can be achieved using neural networks. That is, the hawser tension at a wave height of 4.5m can be lowered to a value comparable to that obtained using conventional control for a wave height of 3.5m. However, in conducting this control, as shown in Figs. 6 and 7, the command of the number of FPSO thruster rotations is overly frequent with respect to economic efficiency. For this reason, improvement of the control logic in consideration of economic efficiency was attempted.

　

 That is, the transverse relative distance between the FPSO and the shuttle tanker in the FPSO fixed coordinate system, as well as its differential value, are added as neurons in the input layer. Moreover, the number of neurons in the middle layer was increased from three points to eight points as shown in Fig. 12. The simulation results obtained from this improvement are shown in Fig.13. The number of thruster revolutions was reduced, while maintaining sufficient position keeping performance.

　

 Positioning performance is considered to have improved, since the amount of information to neural networks increased.

　

 It is thought that optimization can be carried out by means of further improvement in the item selection of the neurons in the input layer and the selection of the number of neurons in the middle layer, or by consideration of the number of thruster revolutions in the neural network structure.

　

 Although good results were obtained in an attempt to improve the dynamic positioning control ability from an economical viewpoint, even more effective neuro-control techniques may also exist. Thus, further examination from various perspectives is required.

　

　

　

 A new positioning control technique for two floating bodies under external forces was developed by adding neuro-control to fundamental PID control. FPSO direction angle is fed back with respect to the position keeping performance during off-loading by a tandem system consisting of an FPSO and a shuttle tanker.

　

 Adoption of neural network control for a dynamic positioning system involving two bodies that consist of an FPSO and a shuttle tanker by computer simulation under various conditions was shown to be valid.

　

 It was confirmed experimentally that the maximum hawser tension under external force conditions of 4.5m significant wave height can be reduced to the same conventional grade value for a condition of 3.5m of wave height by adding the neural control technique to conventional PID control.

　

 The possibility was shown that more economical control performance could be obtained by adding longitudinal relative distance between the FPSO and the shuttle tanker to the input neurons.

　

 In addition, although it was shown that the neural network control method is effective for a tandem offloading system in computer simulation and tank experiments, because neuro-control has substantial potential to improve dynamic positioning performance, it is important to continue the examination of the extension of neural network control through extensive computer simulation.

　

[1] E.Kobayashi, M. Matsuura, K. Daigo, M. Ihara, H. Hirayama and H. Okamoto: "Development of Dynamic Positioning System of Coupled Floating Bodies by Neural Networks", Journal of the Society of Naval Architects of Japan, Vol.190, pp.705-714, 2001(in Japanese)

[2] I. Yamamoto, M. Matsuura, Y. Yamaguchi, A. Tanabe, and K. Shimazaki: "Dynamic Positioning System Based on Nonlinear Programming for Offshore Platform", Proc. 7th ISOPE, Vol.4, pp.632-640, 1997

[3] 1. Yamamoto, M. Matsuura, H. Hirayama, and N. Okamoto: "Dynamic Positioning of Offshore Platform by Advanced Control", OMAE2001 , Rio de Janeiro, Brazil, OFT5102, 2001

　

 Eiichi Kobayashi received the B.E., M. Eng. Sc. degree, Dr. Eng. degrees in 1976, 1978, and 2000, respectively, from the Osaka University, Japan.

　

 Since 1978, he had been a researcher in the field of marine vehicle motion and controllability analysis and coastal environmental analysis at the Nagasaki R&D center of Mitsubishi Heavy Industries, Ltd.

　

 He was appointed in 2002 as a professor at the University of Kobe Mercantile Marine. His current research interests are in the area of maritime technical evaluation, ship control and maneuverability, and maritime and ocean environment.