Fault Diagnosis of Water Lubricated Stern Bearing based on BP Neural Network

Abstract

The water lubricated stern bearing is a special bearings which widely used in ships and other water vessels sail propulsion system,because of its work environment is of great uncertainty,the product is prone to various of failures,once the failure happens,it will directly affect the ship safe navigation and economic benefits. By using the theory and methods of Matlab software based on neural networks,the BP neural network model is constructed to study the work of stern bearing friction coefficient,temperature,noise,vibration and bearing condition characterized by the relationship between the parameters of the time,in order to achieve fault diagnosis for water lubricated stern bearing operating conditions. From Matlab simulation results with the actual situation,the established BP neural network can be a good judge for failed state water lubricated stern bearing,it can send an early warning signal when the bearing failure,will be issued in advance of cattle bearing failure for repair or replacement. By using this method,can effectively improve the use and safety of water lubricated stern bearing.