预成型体渗透率是RTM工艺充模仿真的关键输入参数，本项目开展基于多尺度耦合求解的渗透率的预报方法研究。针对预成型体多尺度结构中随机性和规律性共存的特点，基于势能原理和NURBS曲线曲面理论建立预成型体逼近真实的几何模型。将制造特征的概念引入到RTM预成型体建模中，建立包含微观纤维、细观单胞、宏观特征的预成型体多尺度几何模型。利用Voronoi图方法表达其中的孔隙连通关系，确定渗透率模拟分析的几何边界条件。分析预成型体内部树脂在不同孔隙内的流动和相互耦合机理，基于介观动理学模型和宏观连续模型相结合的多尺度方法建立树脂流动模型。采用粒子法和连续介质模拟法分别对微观和细观树脂流场进行仿真，并采用半通量法实现两者耦合区域的计算，实现预成型体内流场的多尺度耦合模拟和渗透率的预报。采用快速原型方法，制备几何参数规则变化的理想样件进行试验，结合真实预成型体样件的实验，验证和改进渗透率预报方法。

Preform permeability is a key input parameter for mold filling simulation of Resin Transfer Molding(RTM) process, this project propose a permeability prediction approach for RTM preform based on multi-scale coupling solution. Since the structure of the RTM preform preform contains both randomness and regularity, so the principle of potential energy and the theory of NURBS curve and surface are employed to build the geometrical model to guarantee that the model is close to the real structure of preform. The concept of manufacturing feature is introduced into the modelling of RTM preform, the multi-scale geometrical model of preform is established, and the model includes fiber(micro),unit cell(meso) and feature(macro). The Voronoi graph is employed to represent the connective relationship of pores in the preform, which is used to identify the geometric boundary conditions for permeability numerical analysis. The resin flow fields in different pores and their coupling mechanism are analysed, the resin flow models are built based on the multi-scale method which combines the mesoscopic dynamics and macro continuous models.The micro and meso flow fields are simulated by particle method and continuum method, respectively,the coupling area of the two fields is hanlded by the half-flux method, so the multi-scale coupling simulation of the resin flow is conducted and the permeability can be predicted.The ideal preform samples with adjustable geometrical parameters are manufactured by Rapid Prototyping(RP) method to conduct the filling experiment, combined with the real preform flow to validate the permeability prediction model and to improve the predicted permeability tensor.