Thuy Phap THAI

Study on Normalization of Operating Parameters of Shield Behavior with Shield Dimensions

Mitsutaka SUGIMOTO

The shield tunneling method, as a construction technology for urban tunnels, was initially used to minimize interference with road traffic and to prevent construction pollution such as ground subsidence, noise, and vibration when using the open-cut method. In recent years, the peripheral technologies for controlling the behavior of shield machines, hardware, have been developed through advances in measurement technology, improvements in computer calculation speed and the amount of data that can be handled, the spread of the Internet, the application of robust control, and the development of artificial intelligence technology. 
The shield machine is controlled and operated by an automatic tunneling system. However, there are still many unanswered questions about soil properties related to shield excavation, external forces acting on the shield machine, and loads during construction. cutters (length, range). These systems are based on a number of empirical relationships and have no theoretical background. 
In order to solve these problems, it is necessary to apply AI to the field measurement data captured by the computer in real time. This is because in situations where large amounts of data need to be collected and processed to show many possible scenarios on the job site, engineers can use AI to analyze the data and provide results thanks to the sheer volume. There is a large amount of data that has been collected before. Typically, the term "artificial intelligence" is often used to describe a machine (or computer) that can mimic "cognitive" functions that humans often associate with the mind, such as "learning" and "problem solving".
In this study, we derive the relationship between parameters (input values) and behavior (output values) from the results of existing simulations (2020.3) and analysis using shield machines at the Haneda airport intersection and Outer Ring Road sites, and investigate methods for creating AI rules. The parametric study of the shield machine behavior is conducted to understand the features that should be standardized in the shield machine operation support AI. Before conducting the parametric study, it is necessary to make a default case. The reason for this is that the data obtained from the analysis should be close to the planned data when creating the default case, so that the data obtained by future AI will be more trustworthy.