Prediction of topside displacement of offshore platforms based on multi-source data fusion with multilayer perceptron neural network

The top displacement of offshore platforms serves as a critical monitoring parameter for ensuring structural integrity and safety. Given the intricate and challenging nature of these offshore structures, traditional direct displacement measurement methods often encounter practical difficulties and may yield substantial discrepancies in top displacement values. Consequently, alternative indirect measurement techniques, such as strain and acceleration sensing, are commonly employed to gather valuable structural modal information. Nonetheless, the process of deriving precise structural displacement data from these indirect sources can be complex and comes with inherent limitations. In response to these challenges, this paper introduces an approach that leverages the capabilities of a multilayer perceptron (MLP) neural network. By integrating strain and acceleration data from multiple measurable points into the neural network model, the approach facilitates the precise and reliable reconstruction of the top displacement in jacket platforms. To validate the effectiveness and accuracy of this proposed method, comprehensive testing procedures are employed, including finite-element numerical simulations and experimental investigations conducted under various loading conditions. These conditions encompass Gaussian white noise, sinusoidal excitation, and shock load scenarios. The results confirm the feasibility of the proposed method, showcasing its potential to overcome the limitations associated with conventional measurement approaches.