Intelligent Fuzzy Inference-Based Decision-Support System for Marine Main Diesel Engines
TL;DR
The results show that the proposed approach effectively approximates expert reasoning, accounts for nonlinearities and uncertainties, and provides a basis for improving the efficiency of predictive maintenance, optimizing operating modes, and improving the overall operating efficiency of marine diesel engines.
Abstract
The subject of the study is the development of an intelligent decision support system based on fuzzy logic for marine diesel engines, aimed at improving diagnostic accuracy and operational reliability under conditions of incomplete, uncertain, or fragmented data. The relevance of the study is determined by the increasing complexity of parameters monitored by shipboard automation systems and the stringent requirements for safe and efficient engine operation, which create significant difficulties for shipboard specialists in real-time information analysis. The study examines the application of fuzzy logic, production rules, intelligent data analysis methods, and expert knowledge bases to assess the technical condition of marine diesel engines, detect faults at the individual-cylinder level, determine permissible operating modes with due regard to technical and navigational conditions, predict gradually developing faults, and prioritize maintenance activities. The architecture of an intelligent decision support system is developed to provide a transition from conventional monitoring to intelligent information-oriented control in human–machine systems. A fault and troubleshooting database and a knowledge base formalizing production rules using the concept of deviations are considered. A fuzzy inference procedure is implemented, including fuzzification of input parameters obtained from engine monitoring systems, processing using the Mamdani–Zadeh method, and defuzzification of the results to obtain real-time assessments of engine condition. The system provides a graphical interface displaying parameter deviations, probable faults, and operational recommendations. The results show that the proposed approach effectively approximates expert reasoning, accounts for nonlinearities and uncertainties, and provides a basis for improving the efficiency of predictive maintenance, optimizing operating modes, and improving the overall operating efficiency of marine diesel engines.