Identification on Wear Mode for Marine Diesel Engine Based on Evidential Reasoning Rule

To diagnose wear faults of marine diesel engines with uncertain and incomplete information, an evidential reasoning (ER) rule-based model was proposed to identify wear particles of marine diesel engines by using two dimensional (2-D) and three dimensional (3-D) characteristics of particles, which was further used as the basis of identifying wear modes of diesel engines. An oil monitoring system was installed near an engine oil sump to collect oil samples, and the 2-D and 3-D characteristics of wear particles were extracted from the oil samples. Referential points of every characteristic were determined by k-means clustering, and the belief degree distribution of each piece of evidence was acquired by calculating samples’ similarity distribution about referential points and normalizing the likelihoods. ER rule was used to combine multiple pieces of evidence, of which reliability and importance were considered. Five-fold cross-validation and various models comparison indicated that the ER rule-based model can well identify wear particles from marine diesel engines with uncertain and incomplete information, and then the wear mode of marine diesel engines can be determined with the particle type given by the model. The model proved to be simple, accurate, objective and reliable.