HETEROGENEOUS DATA AND PROBABILISTIC SYSTEM MODEL ANALYSES FOR ENHANCED SITUATIONAL AWARENESS AND RESILIENCE OF CRITICAL INFRASTRUCTURE SYSTEMS
MetadataShow full item record
The protection and resilience of critical infrastructure systems (CIS) are essential for public safety in daily operations and times of crisis and for community preparedness to hazard events. Increasing situational awareness and resilience of CIS includes both comprehensive monitoring of CIS and their surroundings, as well as evaluating CIS behaviors in changing conditions and with different system configurations. Two frameworks for increasing the monitoring capabilities of CIS are presented. The proposed frameworks are (1) a process for classifying social media big data for monitoring CIS and hazard events and (2) a framework for integrating heterogeneous data sources, including social media, using Bayesian inference to update prior probabilities of event occurrence. Applications of both frameworks are presented, including building and evaluating text-based machine learning classifiers for identifying CIS damages and integrating disparate data sources to estimate hazards and CIS damages. Probabilistic analyses of CIS vulnerabilities with varying system parameters and topologies are also presented. In a water network, the impact of varying parameters on component performance is evaluated. In multiple, small-size water networks, the impacts of system topology are assessed to identify characteristics of more resilient networks. This body of work contributes insights and methods for monitoring CIS and assessing their performance. Integrating heterogeneous data sources increases situational awareness of CIS, especially during or after failure events, and evaluating the sensitivity of CIS outcomes to changes in the network facilitates decisions for CIS investments and emergency response.