Research On Rough Set And Application On Tunnel Damage Prediction

Data mining is a technique that aims to analyze and understand large source data and reveal knowledge hidden in the data. It has been viewed as an important evolution in information processing. In some aspects rough set overcomes many deficiencies in traditional data analysis, so it has been the most important data mining tool that have been widely used.This paper make a detail study of attribute reduct algorithms and summarizes the characters of main algorithms. Especially a new reduct method which combines the discernibility matrix and search way of heuristic algorithm is introduced. First when computing the discernibility matrix, it select attribute randomly from current term and add it to reduct set which can obtain super reduct set, since there's no need to save the matrix which overcome the extensity difficulty; then eliminate irrelevant attribute by searching super set according to the way of heuristic algorithm. So this method guarantee it can get reduct set and reduce search space. Through data experiment the advantages and validity are proved.Prevention control projecton tunnel damage is so complicated that the method adopted presently which depends on manual basis lacks of systematism. After study on background knowledge about tunnel and damage, this paper employs techniques of data mining on historical tunnel data, so the inherent law can be excavated and the damage rank can be predicted. The techniques rangs from data pretreatment to classification based on rough set and decision tree. Additionally the paper rectify and optimize the calssfication model according to experiment on tunnel data of Chengdu Railway Bureau. The result shows the final model is successful and can obtain accuracy up to 70%. The rules generated can help expert find the inherent reasons of tunnel damage. Finally we also give some advices to railway departments about damage prevention and cure.