Research On Dynamic Positioning Strategy Of Semisubmersible Offshore Platform Based On Predictive Control

Land resources is almost exhausted by the continuous development of mankind. With the continuous economic development, exploration of the resources in the ocean is not just on paper, but an inevitable trend. Difficulty of developing ocean is much higher than development of the land due to the marine environments, thus mankind's exploitation is currently confined to coastal ocean. Because of that most of the marine resources is stored in the deep sea the development of the deep sea is the most important thing in the next few decades. Hence the importance of the development tool semi-submersible offshore platform for ocean seems to be self-evident. As a result, to study on dynamic positioning system(DPS) is clearly obtained seriously. In the past, mooring system is often used as ocean platform in shallow waters, but it has a completely different with working principle with DPS, the latter one is an active system with working principle with DPS, the latter one is an active system of which the working system is more advanced, using control commands send by dynamic positioning unit platform installed on its own platform system to make control orders in accordance with the environmental interference in order to be able to meet the requirements of maritime operations.This thesis studies on a semi-submersible offshore platform, designed the controller in light of the actual control object combined with the popular intelligent control algorithm at present, thereby improving the performance of the DPS.Semi-submersible offshore platform, is a nonlinear system of large delay and large inertia, the results of the current study on which remains difficult for accurate modeling of such a subject. In this regard, this thesis introduced the Least Squares Support Vector Machine(LSSVM) to identify the platform, which is able to be good at the model identification due to the characteristics of LSSVM. Since the conventional Support Vector Machine(SVM) can be inevitably problematic on curse of dimensionality in engineer controlling. LSSVM can effectively overcome this problem, and further to improve the stability of the system, therefore it will be appropriate here to use the generalized predictive control(GPC) as the prediction model.This thesis firstly described in detail the basic algorithms of GPC, and made an emulation with GPC as the control strategy. According to the characteristics of GPC based on prediction model, control strategy based on online GPC of LSSVM has been proposed. The result after the comparing study between optimization of GPC algorithm based on online LSSVM and the non-optimized GPC algorithm in the article shows that, GPC algorithm based on online LSSVM has better robustness and stability than the conventional non-optimized GPC algorithm.The simulation of GPC algorithm based on online LSSVM as the research object shows that, the controller designed according to the offshore platform for the control object shown good control effect of good robustness and stability, of practical effect to the reality application.