Hydrodynamic Performance Analysis Of Savonius-type Impeller Based On Multi-objective Optimization

The Savonius-type(S-type)turbine has the advantages of simple structure,no dependence on flow direction,and good self-starting characteristics.As a vertical axis drag turbine,it relies on the resistance of fluid flow to generate power and has a certain wave energy dissipation capability while continuously capturing energy.It can be integrated with offshore devices such as floating breakwaters as a key component for water energy conversion,and has good development prospects.In order to improve the energy capture efficiency(wave energy conversion efficiency)and wave dissipation efficiency of the S-type turbine,the concave and convex surface shapes of its blades were independently designed,and a guide vane structure was added to create a new MBC-type turbine(Modified-Bach-circle).A multi-objective optimization method combining a BP neural network-based metamodel with NSGA-Ⅱ algorithm was used to optimize five parameters of the MBC-type turbine,including overlap ratio,gap ratio,inner and outer arc angle,inner arc radius,and guide vane size,based on numerical simulation and experimental testing.Firstly,a data set was established using an improved Latin hypercube sampling and numerical simulation method.Then,a BP neural network was trained to construct a metamodel.Finally,NSGA-Ⅱ genetic algorithm and fuzzy decision method were used to obtain the Pareto optimal solution set and select the structure with the highest weighted capture efficiency for hydrodynamic performance analysis..The results showed that the optimal parameters for the MBC-type turbine obtained by this optimization method were: overlap ratio of 0.130,gap ratio of 0.002,inner and outer arc angle of 4.53°,inner arc radius of 42.53 mm,and guide vane size of 4.67 mm.The performance of the optimal MBC-type,traditional S-type,and MB-type turbines were compared and analyzed under the same wave conditions.Compared with the S-type and MB-type turbines,the optimal MBC-type turbine improved the energy capture efficiency by 55.7% and 35.8%,and the wave dissipation efficiency by 5.4%and 1.8%,respectively.This verifies the feasibility of using a BP neural network combined with NSGA-Ⅱ genetic algorithm for multi-objective optimization of the MBC-type turbine.