Mariniello, Giulio (2023) Artificial Intelligence in Structural Engineering: Use Cases for Safety Management. [Tesi di dottorato]

Anteprima

Testo Mariniello_Giulio_35.pdf Download (7MB) | Anteprima

Abstract

Artifcial intelligence (AI) is the most disruptive technology of recent decades. Notably, its implementation in many engineering fields has already begun to show the ability to improve design and construction methods, data management, and safety. Structural engineering plays a crucial role in managing the safety of structures and infrastructures. Whether at the design stage of a new building or during inspections of an existing viaduct, engineers' skills and knowledge must provide realistic judgments that are capable of ensuring the required safety but, at the same time, are economically, socially, and environmentally sustainable. In this process, artificial intelligence makes it possible to assist structural engineers in performing repetitive tasks or duties that require analyzing large amounts of data. This thesis reports several use cases of AI in performing structural safety management operations throughout the structure's life cycle. The first use case is the design of irregular structures. It is complex to obtain structural solutions with high structural performance for such structures using traditional methodologies. Therefore, this thesis propose a metaheuristic strategy for design optimization, the performance of which are compared by the solutions provided by engineering students. The second use case is the management of structural data flows during construction. Proper and transparent management of material acceptance reports, inspections, and load tests is a guarantee of materials and construction processes. An AI and blockchain-based tool for automatic and transparent data management increments safety and provide more confidence in structures. The third and fourth use cases are related to structure monitoring with dynamic and static data. The dynamic is related to the interpretation of accelerometer data for detecting, localizating, and quantifying structural damage understood as reduced stiffness and plastic hinge formation. The static, instead, relates to the evaluation of prestress loss in prestressed concrete lattice bridges from measurements of innovative pressure sensors. The fifth use case is for supporting field engineers in post-earthquake inspections. Deep learning systems can process photos by identifying damage and validating and supporting engineers' opinions. Finally, the sixth use case is for the use of metaheuristics approaches for scheduling maintenance activities for a portfolio of bridges to minimize the portfolio's carbon footprint while meeting the constraints of safety, cost, and available workforce. A methodology based on implementing the Artificial Intelligence methodology was developed for each use case. The methodology were validated using real-world data, where available or already in the literature, and simulated data using computationally robust techniques.

Downloads

Actions (login required)

Modifica documento