Abnormal Condition Detection And Intelligent Decision Of Shield Machine Process

Shield method is an important approach of tunnel construction, involving the research in areas such as mechanics and electricity, control, project management, etc. The encountered abnormal conditions detection and control of shield process are of great importance and become the research focus in shield construction. Multi-agent system is a hot research topic in artificial intelligence. This dissertation takes typical theoretical and application of multi-agent systems as research background, and focuses on the important abnormal conditions recognition and intelligent decision of shield machine. The main contents are arranged as follows.1. Shield machine construction is related to the geological environment, soil mobility in compartment and operation parameter. According to these characteristics, A three-part multi-agent system is developed including integrated geological estimate, abnormal conditions identification, and shield construction adjustment. This level division helps to interpret the problem and improve the precision of detection.2. In the integrated geological estimate part, one Agent of external geological estimate and another Agent of shield cabin soil estimate are established. A belief rule base inference methodology using the evidential reasoning approach is developed as the reasoning machine. Geological exploration data and construction process parameters are integrated to measure the external geological characteristics. Wavelet packet decomposition method is used to extract useful information so as to estimate the earth pressure cabin soil mobility. This geological estimate part provides the necessary support to the abnormal conditions identification and intelligent decision-making.3. In the abnormal conditions identification part, three kinds of Agent is established to estimate the spewing, cake and subsidence conditions. The generation mechanism of abnormal conditions has been analyzed to determine the characteristics of a variable. Then support vector regression machine is used as inference engine to estimate the likelihood and the extent of abnormal conditions. This part gives an early warning and arranges appropriate treatment means.4,In the shield construction adjustment part, one soil liquidity improvement Agent and another two grouting subsidence control Agents have been established. Condition adjustment mechanism has been analyzed, in which soil improvement methods and surface subsidence management tools have been summed. Belief rule base method is employed again so as to find the adjustment means and MPC method is used to automatically find the optimal control strategy.