Integrated approach to creating a case-based database for diagnosing failures in ship power plants

Modern ship power plants (SPPs) represent highly complex technical systems operating under variable, high-load, and often harsh maritime conditions.These systems are characterized by a high degree of component interdependence, making the identification and prediction of failures a challenging task.Traditional deterministic diagnostic approaches often fail to accurately reflect the stochastic nature of technical failures and cascading effects.This article presents a comprehensive integrated approach to failure diagnostics based on a case-based reasoning (CBR) system.The core of the methodology is a structured case base that unites historical failure data, probabilistic models (including Bayesian networks and Markov processes), and discrete-event simulation tools.The structure of each case is standardized and includes failure descriptions, associated risk levels, diagnostic method references, operational context, and interconnection effects.The database currently includes over 5,000 unique failure scenarios derived from operational logs, maintenance records, and simulated data.To increase diagnostic accuracy and system adaptability, the authors implement dynamic updates and automated optimization of parameter weights using the L-BFGS-B algorithm.Simulation modeling is applied to reproduce rare and cascading failure conditions and to improve the generalization capacity of the system.Numerical experiments demonstrate that this integrated approach achieves a diagnostic accuracy of up to 95%, reduces false positives by 30%, and increases system flexibility in real-time operational contexts.The fusion of CBR with probabilistic and simulation models enables the system not only to diagnose known patterns but also to predict new and atypical failures, accounting for system degradation over time.The result is a knowledge-driven support tool for decision-making in ship power plant operations, significantly enhancing operational safety and maintenance planning.This work has implications for the development of intelligent diagnostic platforms in complex marine and industrial systems.