Resilience-Oriented Importance Analysis And Optimization Methods For Infrastructure Networks

The continuous and smooth operations of infrastructure networks have received considerable concerns from both academia and industry.However,infrastructure networks such as telecommunication,electric power,natural gas,and water distribution systems,are prone to failure caused by natural hazards,malevolent attacks or other types of disruptive events,resulting in severe adverse effects on community security and citizen life.The traditional system safety theory,which studies the possibilities and severities of disruptive events in the view of risk,cannot fulfill the need of national economy.Resilience,known as the ability to resist,survival,adapt,and recover,therefore,becomes a paramount feature of infrastructure networks.Due to the complex mechanisms of occurrence and diffusion,the ability of infrastructure networks resisting,surviving,adapting,and recovering from disruptive events cannot be quantified precisely,and eventually,resilience assessment of infrastructure networks can inevitably produce uncertainty.Hence,it's a significant challenge to conduct resilience importance analysis and optimization of infrastructures under various uncertainties.The dissertation devotes to conducting resilience-based importance analysis and optimization for infrastructure networks accounting of uncertainties associated with disruptive events.The research contributions and highlights are summarized as follows:(1)Development of a new resilience-based component importance measure for multi-state infrastructure networks from the perspective of restoration.The proposed methods can quantify the recoverability of infrastructure networks by defining the concept of minimal recovery path,and can characterize the compositive impact of an individual components' recovery on both the performance improvement and recovery duration of the infrastructure network.The resilience-based component importance measure is then proposed and quantified by a probability distribution,and the Copeland score method is utilized to rank the component importance.The illustrative example shows that the importance rank of an individual component changes with respect to the number of failures.If components possess the multi-state characteristic,it may be more effective,from the resilience perspective,to restore some disrupted components to their immediate states.(2)Development of a framework to tackle the uncertainties associated with the disruptive events to obtain optimal design and recovery strategies for resilient infrastructure networks.The proposed framework can systematically analyze the behaviors of infrastructure networks react to disruptive events from the perspectives of both pre-disaster and post-disaster phases.The uncertainties associated with the disruptive events is formulated into a two-stage mixed-integer stochastic nonlinear programming model,and the L-shaped method is utilized to solve the block-structured model to determine a robust design and recovery strategy.The illustrative example shows that the optimal strategy enables the infrastructure networks to maintain high level of resilience under various disruptive events.Additionally,by modeling the uncertainties associated with the disruptive events,the cost of the infrastructure networks in the reacting phase can be reduced.(3)Development of a novel framework to tackle endogenous uncertainty to obtain optimal design and recovery strategies for resilient infrastructure networks.The proposed framework can handle the problems where decisions at one point in time will have a substantial impact on the uncertainty in the later.The occurrence probabilities of scenarios under endogenous uncertainty are derived,and then plugged into a twostage mixed-integer stochastic nonlinear programming model.The probability chain reformulation and linearization reformulation are utilized to linearize the bilinear terms and multi-linear terms,and the nonlinear model is,then,transformed into linear model.After the linearization,the L-shaped method can be readily used to find the best strategy.The illustrative example shows that the optimal strategy not only enables the infrastructure networks to maintain high level of resilience,but also reduces the expected cost under various disruptive events.Furthermore,the failure possibility,network configuration,and recovery resources can simultaneously affect the network resilience.