A control-guided failure restoration framework for the design of resilient engineering systems

When failures are inevitable, a resilient system is expected to restore ideal performance in a timely manner. The resilience of a system can be improved by enhancing the post-failure restoration ability of the system. In order to determine whether resilience in a system is sufficient towards a certain failure, a set of design parameters and performance equations describing the system behavior are essential in performing a resilience assessment. However, in implicit system applications, one of the main concerns is that there are no clearly defined system equations to describe system performance. To overcome this challenge, this paper presents a control-guided failure restoration (CGFR) framework, which combines dynamic system modeling and resilience analysis. Since there are no clearly defined system equations in implicit systems, the dynamic system modeling in the proposed framework is equipped with an artificial neural network to learn system behaviors. To demonstrate the feasibility of the proposed approach, a power transmission system is employed as a case study. The presented study aims to encourage the development of advanced failure restoration strategies for resilient engineered systems.