Abstract: Aiming at the problem of open excavation adjoining bridge piles of Xianggong station construction of Dalian subway line No. 2, the paper constructed the remote automatic monitoring system, arranged the sensors of static force level, electronic yaw equipments and electronic axial force meter, etc. to monitor the information of construction process. Based on Visual C#. Net platform, an intelligent analysis system is developed combining support vector machine and differential evolution algorithm, three- dimensional mechanical parameters back analysis method and time series forecast method are constructed, the software includes four modules of data operation, feedback analysis, data forecast and visualization display. The method and system are applied to the open excavation of Xianggong station. The multiunit messages are collected, then based on the monitoring data, the elastic parameters of the second layer and the third layer are identified by differential evolutionary algorithm; the maximal relative error between the results of feedback calculation and monitoring data is 8.23%. The supporting schemes in the excavation process are simplified accordingly. Furthermore, the sedimentation of bridge pier QCJ-13-01 is forecasted and alarmed by DE-SVM nonlinear time series model; and the measurement of micro tubular insulating piles is adopted at once, which can control the bridge pier settlement effectively. The method combines advanced hardware technology and software analysis function, providing effect approach for complex underground engineering.

Key words: open excavation, multiunit message, time series prediction, integrate intelligence, support vector machine