| Safety monitoring of the tunnel is one of the most important methods for finding the abnormality or potential failure during the service of the tunnel. In the tunnel construction, some device were built to monitor the stability of the surrounding rock. When the project was finished, the surrounding rock was stable and the main factors were temperature and water pressure. There are a lot of difference between construction and running of the tunnel. Up to now, many tunnels run without any monitoring. So it is significant to study the safety monitoring of the running-tunnel.The data collected from the new monitoring system indicate the increasement of the variable such as displacement. However, the initial status cannot be measured directly by device. And the current status equal to the sum of initial status and the increasement, which decides the safety degree of the tunnel. So it is important to calculate initial status.In this thesis, an on-line monitoring and safety evaluation system for the Yinluan Tunnel, a water divert tunnel with 20 years in service, is studied. Existing monitoring data obtained before 1990 are investigated first. Results shows that reinforced concrete lining of the tunnel is mainly affected by the temperature and the permeate water pressure between the lining and surrounding rock, and the effect of rock pressure is much smaller. Then the initial status is estimated and a safety degree model based on adaptive BP neural network is established.Because there are some faults in the back propagation algorithm, an adapt neural network is built and the rule of selection about the training pattern set is studied in this thesis. |
