The Evolving Paradigm of Reliability Engineering for Complex Systems: A Review from an Uncertainty Control Perspective

Traditional reliability engineering paradigms, originally designed to prevent physical component failures, are facing a fundamental crisis when applied to today’s software-intensive and autonomous systems. In domains like aerospace, critical risks no longer stem solely from the aleatory uncertainty of hardware breakdowns, but increasingly from the deep epistemic uncertainty inherent in complex systematic interactions and non-deterministic algorithms. This paper reviews the historical evolution of reliability engineering, tracing the progression through the Statistical, Physics-of-Failure, and Prognostics Eras. It argues that while these failure-centric frameworks perfected the management of predictable risks, they are structurally inadequate for the “unknown unknowns” of modern complexity. To address this methodological vacuum, this study advocates for an imperative shift towards a fourth paradigm: the Resilience Era. Grounded in the principles of Safety-II, this approach redefines the engineering objective from simply minimizing failure rates to ensuring mission success and functional endurance under uncertainty. The paper introduces uncertainty control (UC) as the strategic successor to uncertainty quantification (UQ), proposing that safety must be architected through behavioral constraints rather than prediction alone. Finally, the paper proposes a new professional identity for the practitioner: the system resilience architect, tasked with designing adaptive architectures that ensure safety in an era of incomplete knowledge.