Structural Vulnerability Assessment Procedure for Large Areas Using Machine Learning and Fuzzy Logic

Abstract

This study introduces a general procedure to analyze the structural vulnerability of large areas applied to an unusual problem. Thus, the vulnerability of a whole neighborhood to two simultaneous extreme events has been analyzed. For this task, the Multi-Criteria Decision Making analysis based on experts' judgment has been an excellent option to be used, but it has some drawbacks. For example, it is difficult to make a judgment for every house, especially when there are uncommon problems. Then, freeware has been used for Machine Learning (ML) methods in order to induce every building's performance. Moreover, in order to have a more accurate result in ML, case-based decision theory has been studied to help experts by creating structural analyses’ examples of vulnerability levels. Finally, the Fuzzy Logic Method has been employed to convert the criteria results into smoother numerical levels and to include the uncertainty of experts' judgments. Therefore, an algorithm that includes the interaction with the experts has been developed. Here, the justification for using all those procedures is shown, and it is also demonstrated that they can be applied with freeware.