Abstract：An algorithm for back analysis of displacement is proposed. The method，regarding the monitoring displacement as a goal，estimates rheological mechanical parameters inversely in terms of residual sum of squares of the behavior predicted by the model and the actual behavior of rock mass under certain conditions using the improved parallel particle swarm algorithm hybridized with FLAC3D. Both EBE method and domain decomposition technology are not used to deal with the calculated zone，while optimization algorithm which designs calculating schemes is parallel in the methodology. The difficulties of sub-domains division and exchange of a great deal of data among personal computers are avoided，so efficiency and velocity must be improved well for back analysis of large-scale geotechnical engineering using numerical method. These are proven by analysis of capacity of the algorithm based on the example of section BK14+599 at assistant tunnel B of Jinping second stage hydropower station. The viscoelasto-plastic parameters of assistant tunnel B of Jinping second stage hydropower station covered by about 1 538 m-thick rock mass，as an example，are obtained inversely using the proposed method and the obtained parameters are used for estimation of mechanical and deformation characteristics of assistant tunnel. The results indicate that：(1) the maximum deformation arises at vault and two side walls，mainly distributes along the two side walls；and (2) the phenomenon of floor heave comes up，and stress concentration comes into being at basement and spandrel of assistant tunnel B. These all agree with field investigation. So attentions must be paid to the occurrence of the above phenomena during the design and excavation of diversion tunnels.
