Abstract: Based on the characteristics of monitoring data during highway tunnel construction, a monitoring quality evaluation method combining reliability, saturation and deletion is proposed in this paper, by which data interpolation, graduation and extension can be achieved using polynomials, exponential smoothing, various regression functions, etc. In this way, data availability can be improved effectively, and the rationality of prediction for the rock stability can be ensured as well. Additionally, a feedback analysis is carried out for the rock parameters by adopting a modified PSO algorithm, and the rock stability was predicted for later construction by using a 3D elastic-plastic damage incremental variable stiffness iteration method. From a simulation of the construction of the Changshan highway tunnel, it is demonstrated that adopting the above method can not only guarantee construction safety, but also guide the adjustment of the construction scheme and support parameters in time.