<![CDATA[Early fire detection in ship cargo holds remains a persistent maritime safety challenge, particularly due to the inadequacy of conventional fixed-threshold sensor systems under highly dynamic onboard environments. This study employs a PRISMA-based systematic review of 20 empirical and simulation-based studies retrieved from IEEE Xplore, Scopus, ScienceDirect, SpringerLink, and Web of Science, covering publications from 2010 to 2025. The review examines how adaptive thresholding systems and Fire Hazard Index (FHI) frameworks can integrate multi-parameter sensor data — including gas concentration, temperature, humidity, oxygen levels, and airflow — to improve detection sensitivity and reduce false alarms in maritime cargo environments. Key findings indicate that hybrid approaches combining probabilistic sensor fusion (Bayesian inference, Dempster–Shafer theory) with machine learning classifiers (Support Vector Machine, Random Forest) achieve false alarm reductions of 30–50% compared to static-threshold systems. However, critical gaps remain: fewer than 20% of reviewed studies were validated under real maritime operational conditions, no standardized FHI framework has been adopted for regulatory compliance, and real-time adaptive learning mechanisms remain computationally constrained for onboard deployment. This review contributes a structured synthesis of adaptive algorithm design, FHI integration principles, and system validation requirements, providing a foundation for developing standardized, maritime-specific fire detection frameworks capable of enhancing safety across diverse vessel operations.]]>
Copyrights © 2026
