Navigating Component Degradation and Systems Integration Across the New Energy Vehicle Lifecycle

Abstract

Accelerating electrification exposes severe vulnerabilities in traditional reliability engineering, where uncoupled single-physics analyses fail to capture complex thermal, mechanical, and electromagnetic interactions in modern vehicles. This review critically assesses three core frontiers: multi-physics wear phenomena spanning battery packs, power conversion hardware, and structural glass; hybrid prognostic models reconciling physics-informed priors with machine learning for health estimation; and downstream workflows embedding failure forecasts into maintenance and parts supply chains. We find that although isolated predictive algorithms perform adequately in lab benchmarks, deploying them at the fleet level is severely hindered by the mismatch between localized component diagnostics and real-world supply-chain bottlenecks. Key roadblocks include parameter estimation bottlenecks under shifting dynamics, sample scarcity for nascent failure types, and volatile logistics schedules. Ultimately, advancing next-generation fleet health depends not merely on marginal algorithmic gains, but on standardized benchmark datasets, risk-aware operational policies, and unified interfaces bridging degradation science with logistics operations.