Uncertainty Quantification of Seismic Structural Systems: the Role of Generalized Information Theory

Abstract eng:
Uncertainty is inherent in the assessment and prediction of the seismic performance of structures and community infrastructure when evaluating risk, hazard mitigation and community vulnerability/resilience. Increasingly, community risk-informed infrastructure decisions need to reflect multiple issues of society, including social, political, economic and cross-disciplinary factors. This breadth introduces challenges regarding information/partial knowledge and disparate characteristics of uncertainty from different sources. Traditionally, probabilistic methods have been employed to systematically treat the uncertainty for structural reliability theory. Although these methods or probabilistic methods can address partial information in the face of uncertainty, they are not the most appropriate or powerful approaches for comprehensive incorporation of broader contexts of uncertainty reflecting expert judgement, imprecision, and possibility and evidence theories. Certainly, subjective probabilities can capture expert judgments, but the combination of conflicting opinions remain a challenge. Limitations in probability theory have often led to a failure to fully understand the implications of the broader aspects of uncertainty in human decision-making incorporating judgement and unpredictability. Methods under the umbrella of Generalized Information Theory (GIT) provide a natural framework for linguistic and imprecise data, as is typical from field evaluations both before and after an earthquake. One particular technique, fuzzy classification, has been explored recently to examine general tendencies of damage to concrete buildings from seismic events. This current paper provides some additional background and analysis of that study, and utilizing actual data based on individual building observations, seeks to uncover the existence of building damage patterns among structural, geotechnical and hazard parameters. Additional methods of generalized uncertainty, such as monotone measures, are discussed in the context of augmenting traditional probability approaches to better predict the behavior of buildings during earthquakes.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket