Multi-objective optimization of heat transfer performance and power consumption of Taylor-Couette flow with elliptical helical slits wall

Heat transfer performance and power consumption of Taylor-Couette flow with helical slit wall are analyzed. Slit number, width, and spacing are selected for multi-objective optimization of heat transfer performance and power consumption. Energy loss within the coaxial cylinder is analyzed using the entropy generation principle. Different Machine learning methods are applied to predict the heat transfer and power consumption of Taylor-Couette flow. A comparison made between the predictive findings of the XGBoost model and other three different models. The XGBoost prediction model for heat transfer and power consumption not only exhibits the highest determination coefficient, but also achieves the lowest mean absolute percentage error, root mean squared error, mean absolute error, which has the best predictive performance. Finally, the NSGA-II algorithm is used to optimize the elliptical helical slit structure, and obtained the Pareto front of the optimized design of the helical slit structure. Comparing results with the original model, the maximum improvement in heat transfer performance is 18.68 % and maximum reduction in power consumption is 15.28 %. In practical design, reasonable slit structure parameters can be selected from the obtained set of optimal parameter solutions based on design requirements.