A Hybrid Structural Building Data Mining System Based On the Data Tracking Through Devices

Abstract

One of several traditional megaprojects is underground construction, given its long building time high building expense and possible risks. In tunnel engineering, trench boring devices are generally used to increase work performance and safety. During the tunnelling process, system has recorded vast volumes of tracking data to ensure building safety. The processing of vast real-time surveillance data also lacks successful techniques, and, in many situations, it must be performed manually that pose possible safety hazards. This paper suggests an approach for hybrid data mining (DM) to automatically process the TBM data for real-time tracking. Three separate DM strategies are merged in order to improve the operation of mining also to help security management. The sequential pattern method is executed to remove connections between TBM parameters in order to give people the expertise needed for an irregular on-site judging. A random forest model is built to identify training data in order to complement knowledge needed for building decision-making system. Finally, neural network models measure the penetration rate (ROP) in order to detect irregular data and to alert early. In the case of a tunnel project in China, the suggested technique was applied, and the findings of the application concluded that the approach offered a reliable and effective way of evaluating TBM protection management data in real time during buildings.