Full-condition operation prediction and parameter control model of ocean thermal energy conversion experimental platform based on improved neural network

To achieve optimal performance of the ocean thermal energy conversion (OTEC) experimental platform under full working conditions, accurate full-range operational control becomes critical. To resolve this challenge, this study proposes the development of predictive and control models for a full-condition operation of the OTEC experimental platform. The model contains the OTEC model, improved neural network (IPSO-IBP) algorithm model, multi-objective optimization and control parameter screening model. Based on the OTEC experimental platform adopting a 50 kW dual-turbine power generation unit with gas lubricated bearings, 213 sets of experimental data are used to investigate the influence of external controllable parameters, i.e., warm/cold seawater inlet temperatures and mass flow rate on the platform's performance. The study addresses the multi-objective optimization challenge of maximum grid-connected power and maximum thermal efficiency under full operating conditions, identifies optimal control parameters for various scenarios, and predicts annual optimal power generation considering seasonal temperature variations in the South China Sea. Results demonstrate that the improved neural network enhances prediction accuracy compared to other models, the prediction precision for grid-connected power, thermal efficiency, and turbine speed improved by 0.5 %, 4.9 %, and 1 %, respectively, with absolute errors ranging within −0.4–0.5 kW, −0.07–0.08 %, and -86–38 rpm. Among adjustable parameters, warm seawater temperature exhibits the most significant influence on system performance, while cold seawater mass flow rate shows minimal impact; through the NSGA-Ⅱ algorithm of optimization and the LINMAP decision-making, the annual optimal power generation in the South China Sea reaches 211626.6 kW∙h. This study establishes theoretical and data-driven foundations for the control of practical OTEC power generation applications.