基于RBFNN重构油膜力的发动机连杆瞬态变形分析

he transient deformation of a connecting-rod including effect of two-dimensional oil film force for the piston pack, was investigated by using a Radial base function neural network(RBFNN). To this end, a RBFNN with three layers was designed to reconstruct the oil film forces calculated from associated lubrication equations. Then, the trained oil film forces by the well-trained RBFNN were incorporated into the dynamic equations for the piston-crankshaft system to achieve forces acting on the connecting-rod. After this, finite element analyses of transient deformation for the the connecting-rod were made through software ANSYS. It is found that the connecting rod’s deformation varies with its rotation. Moreover, its maximum deformation occurs at the time when the piston is before or behind the bottom dead center at compression stroke, which implies the conventional strength check method of the connecting rod is not valid for the rotary connecting-rod. Meanwhile, the validity of the proposed method was demonstrated.