Abstract:  In order to analyze the influence of labyrinth seal parameters on the leakage of a wind power gearbox, the leakage behavior of labyrinth seal was studied by numerical simulation and proxy model. Firstly, the threedimensional flow field model of labyrinth seal was constructed according to the three-dimensional solid model of wind turbine gearbox. Then, based on FLUENT software, the changes of labyrinth seal leakage with inlet and outlet pressure ratio, dynamic viscosity of lubricating fluid, speed of highspeed shaft and low-speed shaft and seal tooth clearance of labyrinth seal were studied by single variable method. Finally, a radial basis neural network (RBF) proxy model was constructed. On the basis of this proxy model, three structural parameters, seal gap, seal cavity height and seal cavity width, were selected as design variables among the factors affecting the labyrinth sealing performance, and the minimum leakage and maximum exit velocity of the labyrinth seal were taken as optimization objectives. Non-inferior hierarchical genetic algorithm (NSGA-Ⅱ) was used to obtain the optimal solution. The results show that the leakage of labyrinth seal is little affected by the speed of two rotating shafts. The leakage is proportional to the ratio of inlet and outlet pressure and seal clearance, and inversely proportional to the dynamic viscosity of lubricating oil. When the parameters corresponding to the optimal solution are obtained, the leakage is reduced by 47% and the maximum exit velocity is reduced by 36%. The numerical simulation element results are consistent with the optimization theoretical calculation results. The results provide a theoretical basis for studying the influence characteristics of labyrinth seal leakage.
Key words: wind turbine; gear case; labyrinth seal; leakage volume; numerical simulation; optimal design; radial basis neural network(RBF) proxy model