<p>System assurance has become essential in modern engineering systems, driven by rapid electrification and digitalization across industries, where the frequency of disruptions continues to rise. In response, research has focused on developing robust operation strategies, including quantitative resilience measures. However, most existing resilience metrics model system recovery process from a limited perspective and often treat systems as monolithic entities, overlooking internal structure information. There remains a gap in system-level resilience metrics tailored to multi-component engineering systems that explicitly incorporate system architecture. To address this, we propose a probabilistic resilience framework that quantifies resilience from three perspectives: recoverability, recovered loss, and recovery speed. Based on this framework, we propose a set of system-level resilience metrics that explicitly account for the design configurations of multi-component systems. The proposed metrics are applied to battery energy storage system design problems using experimental data collected from laboratory tests. The results demonstrate the practical applicability of the proposed resilience metrics in supporting engineering system design and evaluation.</p>