Maritime systems operate in highly dynamic environments where unexpected equipment failures jeopardize safety, reliability and efficiency. Recent advances in AI, machine learning, digital twins and predictive maintenance enable proactive failure prediction and prevention, yet ensuring trustworthy and explainable decisions remains a core challenge for safety‑critical maritime applications. This article reviews the essential AI technologies for explainable failure prediction and proposes a conceptual architecture that supports autonomous or human‑in‑the‑loop corrective actions. The architecture integrates data acquisition, time‑series forecasting, anomaly detection, risk assessment, decision making and explainable AI into a closed‑loop framework, covering the full chain from perception to intervention. Based on these components, we survey existing maritime studies, summarizing their methods, strengths and limitations, and highlight current challenges such as uncertainty, robustness, model generalization, explainability, scarcity of maritime datasets and deployment issues, outlining future research directions toward trustworthy AI‑assisted maritime decision making.
Review