Research on Control of Winch Heave Compensation System Based on Wavelet Neural Network Velocity Prediction

Focusing on an energy-saving winch-type heave compensation system applicable to real working conditions, with the objective of enhancing compensation accuracy, a wavelet neural network was employed for platform velocity prediction, and the prediction results were applied to velocity disturbance compensation control. Initially, the ITTC two-parameter spectrum was utilized to generate wave spectral diagrams under different sea conditions, along with displacement and velocity data of the floating platform’s heave motion. Subsequently, a time-series-based wavelet neural network velocity prediction model was developed, trained, and tested. Comparative analyses were performed on prediction performance differences across varying prediction steps and sea condition levels. Then, the effectiveness of the time-series-based wavelet neural network prediction model was validated through a valve-controlled hydraulic cylinder heave motion simulation system. Experimental results indicated that the wavelet neural network-based velocity prediction method effectively improved the compensation accuracy of the winch-type heave compensation system. Finally, after verifying the effectiveness of the wavelet neural network prediction model based on time series, the compensation performance of the system after adding the velocity prediction module was tested and verified using the winch-type heave compensation simulation test bench built by the research team. After experimental verification, after adding velocity prediction, the compensation accuracy of the system was improved by 19% compared with that without velocity prediction.