Loading...

Enhancing the Resilience of Urban Infrastructures: A Combination of Machine Learning-Based Methods and GIS

Hajbarat, Mohammad Amin | 2024

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 57857 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Haj Kazem Kashani, Hamed
  7. Abstract:
  8. Throughout history, the occurrence of natural disasters and incidents has posed a potential threat to man-made constructions. Advancements in science and technology have led to the development of man-made civil infrastructures, which now exhibit increased complexity in terms of technological sophistication and spatial coverage. Consequently, monitoring and managing their health and risk factors have become more intricate. Urban lifelines, which encompass constructed utilities, transportation systems, telecommunication networks, and critical facilities, hold significant importance for the well-being of citizens. These infrastructure assets are essential for the distribution and flow of services and supplies in urban centers, impacting human livelihoods and economic prosperity and natural hazards present a peril to their existence. It is a proven need to make suitable decision and policies for enhancing their resilience against natural hazards in order to reduce the socioeconomic risks faced by communities. By enhancing the resilience, the initial damage caused by the disaster occurrence will be mitigated and on the other hand, the recovery speed will be accelerated. This research work proposes a novel framework to predict the spatial probability of natural hazard occurrence utilizing a combination of Geographic Information Systems (GIS) tools, Remote Sensing (RS) and Machine Learning (ML) methods. The ultimate objective of this framework is to generate a visual representation of risk and resilience indices within a specified area. This visualization aids experts and decision-makers in formulating optimal development and retrofit strategies. After the framework prepared, in a sample area situated in Mazandaran province, Iran, the probability of landslide occurrence has been predicted using a geospatial machine learning (ML) model. The model achieved an impressive AUC validation score of 98%. Subsequently, the direct and indirect cost-based risk and resilience for this community were quantified. This assessment considered the hazard probability associated with Natural Gas (NG) network assets and their dependent facilities. To enhance risk communication, a novel visualized method was proposed, facilitating a clearer understanding of risk levels by the audience
  9. Keywords:
  10. Machine Learning ; Geographic Information System (GIS) ; Community Resilience ; Resilience Enhancement ; Risk Analysis ; Hazard and Operability Analysis (HAZOP) ; Networked Civil Infrastructures ; Natural Disasters

 Digital Object List

 Bookmark

No TOC