Explainable AI (XAI) for PHM of Industrial Asset: A State-of-The-Art, PRISMA-Compliant Systematic Review

XAI improves industrial equipment monitoring without sacrificing performance: The review found that papers combining Explainable AI with Prognostic and Health Management (PHM) produced "good or excellent results in general," suggesting that making AI systems more transparent doesn't compromise their ability to predict equipment failures or detect problems.

There's a real-world need for explainability in industrial settings: XAI serves a dual purpose in PHM—it both helps complete maintenance tasks and provides explanations for decisions. The review found this is particularly important for diagnostic work and anomaly detection, indicating that industrial practitioners genuinely need to understand why their AI systems make recommendations.

Human involvement in AI-driven maintenance decisions needs more attention: The systematic review identified that "human role, evaluation metrics and uncertainty management are areas requiring further attention," suggesting current AI systems for industrial maintenance don't adequately account for how humans should interact with and validate AI recommendations.

Industrial applications are grounded in real data: Most case studies reviewed used actual industrial data rather than simulated scenarios, indicating that the explainable AI approaches being developed are practical and capable of handling complex, real-world maintenance challenges—not just theoretical exercises.

Better standards for evaluating explainability in maintenance are urgently needed: The review emphasizes that "adequate assessment metrics to cater for PHM need are urgently needed," meaning the field lacks consistent ways to measure whether explanations are actually useful to maintenance professionals.