Research On The Fault Diagnosis For The Shipboard Integrated Electric Propulsion System

The ship integrated electric propulsion system plays an important role in the next generation of advances ships.With the increasing demand for power of propulsion load,ship service load,weapon system and mission system,a stable and reliable ship integrated electric propulsion system is essential to maintain all aspects of ship operations.It has also become an important support to improve the operation efficiency,economy,reliability and survivability of ships.The ship medium voltage direct current(MVDC)integrated electric propulsion system is a kind of tightly coupled and complex network structure.During the ship's navigation,faults may occur in the equipment interior,DC bus or node.The fault usually has a serious impact on the electrical subsystem,thus endangering the normal navigation of the ship.Therefore,it is of great significance to quickly detect and locate different types of faults in the system under different operating conditions to ensure the safe operation of the ship integrated electric propulsion system.In this paper,several key problems in fault diagnosis technology of ship MVDC integrated electric propulsion system are studied:Firstly,the fault diagnosis architecture of ship integrated electric propulsion system is designed.The typical faults and phenomena of key equipment in the system are summarized.The fault tree of ship integrated electric propulsion system is constructed,and the fault diagnosis architecture of fault tree expert system and intelligent fault diagnosis method is designed.We introduce the structure and functions of the ship integrated electric propulsion ystem,and study the operation mechanism of the system.Then,according to the function division of each part of the system,the mathematical model of power generation system,propulsion system,regional load and distribution system are established.It provides the necessary overall structure framework and mathematical model basis for the future research on fault characteristics analysis and fault diagnosis technology of ship MVDC integrated electric propulsion system.Secondly,the fault feature extraction technology in the intelligent fault diagnosis method of ship integrated electric propulsion system is studied.In order to refine fault information,eliminate interference components and reduce fault feature dimension,a fault feature extraction method based on Noise Assisted Multiple Empirical Mode Decomposition(NA-MEMD)and energy moment theory is designed.NA-MEMD is used to preprocess the fault data,and the key information in the decomposition results is extracted by energy moment.According to the characteristics of the fault signal,the velocity modified-mutation particle swarm optimization(PSO)-NA-MEMD method is proposed.The idea of mutation is introduced in this method to expand the search range of particles,and the adaptive selection of white noise amplitude parameters is realized.Numerical simulation and measured data analysis of the fault feature extraction method of the ship MVDC integrated electric propulsion system are carried out to verify the effectiveness of the fault feature extraction method proposed in this paper.Then,the electrical fault diagnosis and location technology in ship integrated electric propulsion system is studied.This section describes the typical electrical faults in ship MVDC integrated electric propulsion system.We discussed the causes of the faults,the phenomena after the faults and the influence on the system,then the electrical fault is simulated.In order to solve the problem of training gradient deviation caused by unbalanced number of normal samples and fault samples in training samples,a fault diagnosis method based on empirical mode decomposition and weighted iterative LightGBM is proposed.In this method,the classification confidence evaluation method is designed,and the weight coefficient of training samples is introduced to realize the diagnosis of AC fault and DC fault of ship MVDC integrated electric propulsion system under unbalanced samples.Then,the research on electrical fault diagnosis and location technology of ship MVDC integrated electric propulsion system under multiple operating conditions is carried out.A framework integrating electrical fault diagnosis and fault location is proposed.Only by detecting and analyzing the DC bus voltage,the types and locations of AC and DC faults in the system are determined.This method reduces the influence area of system fluctuation on fault diagnosis and location methods.This section also studies the influence of system operating conditions and parameter perturbation on fault diagnosis and location method,and designs an electrical fault diagnosis and location method with strong generalization ability.The effectiveness of the proposed electrical fault diagnosis and location method for ship MVDC integrated electric propulsion system under multiple operating conditions are studied.The validating simulation results show that the proposed method has strong generalization ability and effectiveness.Finally,the mechanical fault diagnosis technology in ship integrated electric propulsion system is proposed.The typical mechanical fault characteristics and phenomena of ship integrated electric propulsion system for ships are studied.Combined with the characteristics and commonness of mechanical fault,the idea of mechanical fault diagnosis for ship integrated electric propulsion system is designed.Considering the problem that it is difficult to obtain a large number of fault samples in practical engineering,which results in insufficient model training,a mechanical fault classifier based on improved functional fuzzy neural network is proposed.Combining with the fault feature extraction method,the network structure of functional connection network and fuzzy neural network is designed.In this network structure,trigonometric function is introduced as the basis function to complete the function expansion of input data,and the state layer is constructed to increase the learning efficiency of limited training samples.The network realizes accurate and rapid diagnosis of mechanical faults in the case of small samples.The typical faults of ship propulsion motor and gas turbine are taken as examples to verify the effectiveness of the proposed ship integrated electric propulsion system fault feature extraction method and mechanical fault diagnosis method.