A system of systems flexibility framework: A method for evaluating designs that are subjected to disruptions

Systems of systems;Modeling and simulation;Level of repair analysis;Genetic algorithm;Graph theory;Discrete event simulation;Network modeling;Flexibility;Optimization

Warshawsky, David ; Mavris, Dimitri Aerospace Engineering Cooksey, Daniel Schrage, Daniel Kennedy, Graeme Salmon, John ; Mavris, Dimitri

University:Georgia Institute of Technology

Department:Aerospace Engineering

关键词: Systems of systems;    Modeling and simulation;    Level of repair analysis;    Genetic algorithm;    Graph theory;    Discrete event simulation;    Network modeling;    Flexibility;    Optimization;   

Others  :  https://smartech.gatech.edu/bitstream/1853/54277/1/WARSHAWSKY-DISSERTATION-2015.pdf

美国|英语

来源: SMARTech Repository

【 摘 要 】

As systems become more interconnected, the focus of engineering design must shiftto include consideration for systems of systems (SoS) e ects. As the focus shifts fromsingular systems to systems of systems, so too must the focus shift from performancebased analysis to an evaluation method that accounts for the tendency of such largescale systems to far outlive their original operational environments and continuallyevolve in order to adapt to the changes. It is nearly impossible to predict the natureof these changes, therefore therst focus of this thesis is the measurement ofthe exibility of the SoS and its ability to evolve and adapt. Flexibility is measuredusing a combination of network theory and a discrete event simulation, therefore,the second focus is the development of a simulation environment that can also measurethe system's performance for baseline comparisons. The results indicate thatsimulated exibility is related to the performance and cost of the SoS and is worthmeasuring during the design process. The third focus of this thesis is to reduce thecomputational costs of SoS design evaluation by developing heuristics for exibility.This was done by developing a network model to correspond with the discrete eventsimulation and evaluating network properties using graph theory. It was shown thatthe network properties can correlate with simulated exibility. In such cases it wasshown that the heuristics could be used in connection with an evolutionary algorithmto rapidly search the design space for good solutions. The entire methodology wasdemonstrated on a multi-platform maintenance planning problem in connection withthe Navy Hardware Open System Technologies initiative.