The surrounding rocks of Shawan Extralong Tunnel on Test Bid Section 2 of WeiyuanWudu Highway in Gansu are weak and broken, which is softening in water and with low strength. Hence, the ABAQUS numerical calculation and orthogonal design method are used to analyze the parameter sensitivity so as to learn the influence of each parameter on the deformation of surrounding rock. A PSOBP parameter inversion analysis method is established by BP neural network algorithm based on particle swarm optimization; and back analysis is programmed. The displacement, stress and plastic zone distribution of surrounding rock and supporting structure are analyzed by numerical simulation based on the inversion parameters. The study results show that: (1) The factors affecting the displacement of the surrounding rock of Shawan Tunnel are cohesive c, Poisson′s ratio μ, internal friction angle φ and elastic modulus E; (2) The PSOBP parameter inversion analysis method is an effective inverse analysis method with high accuracy.