Data Mining For Bridge's Monitoring Data Based On Netural Network

Along with the bridge accidents have increased year after year, the Government and the Engineering are paying more attention to the forgetful monitoring for the bridge. The bridge is a very complicated system, it's deterioration under the normal circumstances is a slow process. The normal monitoring system of bridge will produce much data each day, but the aging and injury of the bridge often occur during a long time, so the data of collection is stupendously. But the commonly system don't have capacity to processing the data (is general at GB level), so it can't find the problems which maybe exist in the bridge through the historical data that much valuable data is abandoned.This Dissertation tries to use the Data Mining technology based on neural network mining the bridge monitoring data, finds a data analytic model of the data, obtains the change(fluctuate) rule of a bridge parameter, mines the relationship between the bridge parameter. In fact the study also has practical significance on analyzes the monitoring data of the big buildings such as skyscrapers,stadiums,dams,airports etc.The main content in the dissertation is based on the Data Mining technology, combined the status of the bridge monitoring system, and proposed a solution which used the capacity of the Data Mining in processing large data sets to analyze the data generated by bridge monitoring system, used the Data Mining technology based on neural network to analyze the data, processed the large amounts of data that generated by the bridge monitoring system for a long period, built up kinds of Data Mining Model. The model are:①C luster model, mainly used for monitoring the unconventionality of the bridge data, stipulate for the data.②Associate model, mainly used for find out the associate rule between the bridge architecture and environmental parameters.③time series analysis model, mainly used for observed the change trends of the bridge monitoring data, compared the change trends of kinds of parameters and predicted the value of the parameters. Through contrast the time series figure of the parameters, we can also find the relationship between parameters. Using kinds of the established Data Mining Models, we can judge and recognize the current bridge's healthy status, mine the useful information hidden in the data to enhance the bridge's health.The dissertation research results can greatly improve the automation and efficiency of the bridge's health status assessment. But the analysis of the bridge's monitoring data is a complex system, which involves many aspects; meanwhile how to correctly judge the bridge's healthy status in time, and improve the monitoring accurate for bridge's healthy status is also a problem to research. So there needs to do a lot of work to put the bridge's monitoring data mining system based on neural network into use.