Research On Knowledge Discovery Of Metro Construction Safety Based On Multi-Granularity

The factors involved in the safety risk of metro construction project are quite complicated, and the occurrence of construction accident are co-determined by geology setting, hydrological condition, construction method, shield parameters and so on. The disaster evolution law is pretty complicated, safety early-warning is greatly dependent on the expects' experiential knowledge. Meanwhile the monitoring data while doing large-scale excavation in metro construction is massive, multi-source heterogeneous, uncertain and complicated in dynamic effect. With the gradually popularization of engineering information and network, the contradiction of rich data and poor knowledge becomes increasingly prominent, the shortage of traditional manual processing method enormously influences the accuracy and timeliness of safety risk perception and decision control of metro construction. Around the above-mentioned question, this paper uses the Cloud Model, combined with some monitoring data in practical case, and takes the conception classification of risking factors. Based on the above results, attributes reduction and decision rules are obtained by Rough Set theory, finally the fuzzy control technology is used to carry out the reasoning and decision analysis, which providing effective support to safety management of metro construction.In this thesis, combined with the relevant documents of the safety control theory and knowledge discovery method and existing engineering practical experience, the information granulation is introduced, which has strong adaptability to complicated data. The use of soft division function of Gray Model theory makes it true to gain the concept division of the several risking factors in metro construction and obtains the discretization result of continuous attributes in the decision table. By using the Rough Set theory, the method of multi-granularity knowledge attributes reduction in complex data is obtained, which makes the reduction result the same classification ability to the decision attribute, while getting the most reduction group of knowledge space. And this strengthens the processing and recognition ability of large scale complex data in metro construction environment. Through the design of fuzzy controller of the metro construction safety reasoning and decision-making system, it is proved feasible to realize the reasoning and control of metro construction, combined with the rules acquired above.