Fault Diagnosis Of Marine Power System Based On Hopfield Neural Network

The ship's power grid is one of the most important power networks for ships,and ships with larger and more complex power networks are being built as the economy grows.When the ship is in the process of voyage,once the power system fails,the local mechanical failure is caused,and the entire power system is paralyzed,and even the collision of the ship may occur.Therefore,the power system of the ship during navigation is not allowed to have major power failures.If the danger occurs,the fault should be discovered and dealt with in time.The premise of quickly repairing the fault is to identify the type of fault,the location and cause of the fault.This study uses a discrete Hopfield neural network(DHNN)for ship network fault diagnosis.Taking the simulated ship power station of the School of Shipping and Ship Engineering of Chongqing Jiaotong University as the prototype of the power system for this study,the fault of the conventional power grid can be confirmed by observing the action of the relay switch,but if the switch refuses or the relay fails.In the case of a situation,it is difficult to visually detect the fault condition,and the diagnosis of the fault requires the maintenance personnel to take time and effort to conduct the investigation.This can even lead to an accident if the ship is in a special situation.Therefore,this study mainly solves the rapid identification of ship network fault types,and classifies the common faults of ship power grid into three categories: conventional faults,protection devices,faults,and combined faults.The purpose is to quickly identify these three types of faults.The specific idea is to collect the whole grid switch and protection relay status(passing 1,breaking to 0)for encoding,then classifying the fault type and obtaining the corresponding switch and relay status according to the circuit analysis method and fault tree fault analysis method.As an input sample of the discrete Hopfield neural network.With the characteristics of the associative memory of the discrete Hopfield neural network,when the above three types of faults occur again,the diagnostic system can well identify which fault condition the fault of the grid belongs to and make a response strategy.Finally,the simulation tests of conventional faults,switch relays,and combined faults show that the discrete Hopfield neural network can effectively identify the above three types of faults,and a simple ship grid fault diagnosis system is written by using SQL Server database and MATLAB.Increase its practicality.