Prediction Algorithm And Experiment Research On Active Heave Compensation Of Offshore Crane

Offshore crane with active heave compensation system(AHCS) is an intelligent ocean engineering equipment, which can compensate the heaving disturbance caused by environmental factors such as ocean wind, wave and flow, thereby ensuring the the safety of offshore lifting. However,in practice, time-delay between the sensors and actuating mechanism makes poor control quality and decreases the compensation precision of the control system. If we can predict the short-term heave values accurately, and accord to the predicted values, control unit will output the controlling commands to skip the dead times in the system, so as to improve the compensation accuracy and stability of the AHCS. Therefore, studying in the heave motion prediction algorithm has essential meaning in theory and practical value for the development of AHCS.The research is carried out by theory analysis, computer simulation and test analysis in the dissertation. The main problems that will be solved in this paper are as the following:First of all, this research deeply analyze the intrinsic characteristics of the offshore crane’s heave motion time series, the chaotic characteristic parameters of the time series indicated the ship heave motion time series have chaotic characteristics, established the Least Squares Support Vector Machines(LSSVM) multi-step prediction model after reconstructing the heave motion sequence by using phase space reconstruction technique, predicted the heave motion on the different direction angle of waves. The simulation results show that compared with the single LSSVM forecasting model, forecasting model based on the chaos theory and LSSVM, fully exploited the characteristics and regularity of the offshore crane’s heave motion time series, which makes the prediction results more accurate.Secondly, according to the principle and basic structure of the Extreme Learning Machine(ELM) and online sequential extreme learning machine(OSELM), in view of the instability of the COSELM forecasting model simulation results, proposed an improved algorithm based on OSELM——Enhanced Search Online Sequential Extreme Learning Machine(ES-OSELM), It can select optimum hidden layer nodes of extreme learning machine by error minimization method, applied the model to forecast the heave values 20 seconds ahead, simulation result shows that the CES- OSELM has a better stability and prediction accuracy than COSELM model.Finally, in order to examine the accuracy and timeliness of predict algorithm above, designed a semi-Physical simulation system of offshore crane’s motion online prediction, which including a three-degrees-freedom simulation platform 、 hoist mechanism、control unit of simulation platform、ship motion measure and calculation. According to the ship’s movement data, the control unit drives the experimental platform to simulate the ship’s motion, movement forecast unit receives the date of offshore crane’s movement in real time, then model and forecast. The experimental results indicate that motion prediction accuracy of next 5s in CES-OSELM model is better than CLSSVM model. This forecast algorithm have great applied value for improving the control performance and stability of offshore crane’s AHCS, and are worth widely popularizing.