Research On Predictive Maintenance Technology Of Diesel Turbocharger Based On Machine Learning

As a rotating machinery,turbocharger is one of the key equipment of Marine diesel engine.Its structure is complex,and it usually works in bad environment.When failure occurs,it has great influence on diesel engine.With the development of social economy and technology,the automation level of ship equipment is getting higher and higher,and the need for continuous work of equipment is getting stronger and stronger.Equipment shutdown caused by sudden failure of ship equipment will reduce navigation efficiency,affect the navigation safety of ship,and threaten the safety of life and property of crew and passengers.The turbocharger system is complex,the failure will affect the running state of the diesel engine,cause the service life of the diesel engine is reduced,and threaten the navigation safety of the ship.Therefore,it is of great significance to study predictive maintenance technology of Marine diesel turbocharger.This paper firstly analyzes the typical faults of supercharger.The working principle and structure of turbocharger are introduced in detail,and the typical faults of turbocharger such as surge vibration,pressure anomaly,noise,oil leakage and overtemperature are analyzed.The typical fault tree of turbocharger is established,and the fault characterization analysis of these faults is further carried out.The predictive maintenance technology research based on the vibration data of turbocharger is proposed.Then the development and realization of predictive maintenance technology and the maintenance status of Marine diesel turbocharger are described by referring to documents.This paper introduces the basic concept of predictive maintenance technology in detail,makes an in-depth study of predictive maintenance technology based on machine learning,expounds the development of machine learning algorithm in predictive maintenance technology,and uses Pearson correlation coefficient and LSTM algorithm to build the predictive maintenance algorithm model framework of Marine diesel turbocharger.Finally,the predictive maintenance technology of turbocharger is researched and verified.The predictive maintenance model based on Pearson correlation coefficient and LSTM algorithm was realized by algorithm programming.The feasibility of the algorithm model was verified by analyzing the sailing vibration data of two real turbocharger ships of the same model of a diesel engine.