Data Driven Fault Diagnosis And Maintenance Decision For Marine Bearings

Bearing is an important component of the ship systems and plays an important role in the safe and reliable operation of the ship.Therefore,it is very important to carry out research on fault diagnosis and maintenance decision of marine bearings.Nowadays,it is possible to collect vast amounts of data that reflect changes in the health status of marine bearings relying on condition monitoring method centered on sensor technology.This paper analyzes the research background and status quo of bearing fault diagnosis and maintenance decision-making,adopts a data-driven approach,and thoroughly studies the fault diagnosis and maintenance decision methods for marine bearings.The main research content includes:Firstly,taking full advantage of the merits of wavelet analysis in weak signal processing,the method of early weak symptom recognition for marine bearings is set forth,which is based on multi-scale wavelet packet transform.Combining the advantages of deep convolutional neural network in image recognition,a fault diagnosis method marine bearing based on deep convolutional neural network and S-transform is proposed.Secondly,the large-scale monitoring signals of bearings are preprocessed,and the features are extracted from different domains,which reflect the changes of bearing status,including time domain,frequency domain and time-frequency domain.On this basis,the correlation features and monotonicity analysis are performed on the extracted features to achieve feature selection.The principal component analysis method is used to fuse the selected features.Then,the bearing multi-objective maintenance decision method is studied.A degeneration model based on the Wiener process is established.The particle estimation algorithm and Monte Carlo algorithm are used to realize parameter estimation and degradation prediction,revealing the bearing reliability trend.A bearing life prediction method based on empirical Bayesian algorithm is proposed.Based on the degenerate modeling and life prediction,considering the bearing reliability,cost rate and availability,a multi-objective model of marine bearing maintenance decision is established.Finally,an experiment of studying accelerated bearing life was carried out.Design and build an experimental platform and place different sensors such as vibration,temperature,and current to obtain massive bearing monitoring data.The proposed method was used to analyze the experimental monitoring data,which verified the feasibility and practicality of the proposed method.