Research On Fault Diagnosis Method For Marine Current Turbines Under Variable Operation Conditions

The exhaustion of non-renewable energy has caused a series of problems such as global warming.The "Dual-Carbon Goal" is a major strategic move China has made in the face of environmental problems,and new energy development is a key part of the "Dual-Carbon Action".The marine current turbine(MCT)can convert marine current energy into electrical energy,but due to their complex working environment,it is difficult to diagnose them effectively when an MCT fault occurs.Therefore,it is important to study efficient fault diagnosis method.This thesis is funded by the project "Research on fault diagnosis method of MCT based on multi-domain characteristics"(Project No.61673260),and takes the MCT as the research object.The research is based on the current signal-based fault diagnosis method of MCT.The main work of the thesis is as follows:(1)MCTs are located in a complex submarine environment,where offshore winds and surges occur randomly.To address the problems that the fault model of MCTs cannot be accurately established by the working conditions and its signal features are difficult to be extracted accurately,this thesis models and characterises MCTs and their common faults,and extracts the fault features based on the stator current signal.Further,an Extended Concordia Transform(ECT)based fault feature extraction method is proposed to enhance the effect of fault features by using first order derivative to improve the modulus signal.The validity of the method is verified based on a prototype testbed.(2)Due to the influence of celestial motion,the current cycle fluctuates in different flow velocity regions,which makes the MCT switch back and forth between different operating conditions.To address the problem that it is difficult to detect the blade impact fault under complex marine conditions,this thesis proposes an Envelope Geometric K-means cluster(EGK-means)based method to classify the MCT operating conditions,and finally establishes a multi-segment Principal Component Analysis(PCA)fault analysis based on the operating condition information.And the process statistics is used as the fault detection threshold to achieve the impact fault detection.(3)Due to the influence of random offshore winds and surges,the seawater has a swelling effect.To address the problem of drifting and blurring of blade biofouling fault features,an Adaptive Proportional Frequency Sampling(APFS)method is proposed to combat the problem of poor fault characterisation capability by adaptively adjusting the sampling frequency to update the modulus signal.The updated modulus frequency domain signal is meet the requirements of detectable fault features.In term to the abundant of the frequency domain signal variables and the complexity of the calculation,the redundant variables are eliminated by transconversion amplitude screening to initially reduce the feature dimensionality,and finally the fault detection of the attachment is achieved by the PCA model.(4)The harsh subsea environment and random irregular fluctuating loads make the mechanical and electrical components of the MCTs prone to various kinds of failures to affect the performance of the whole machine.To address the problem that it is difficult to accurately diagnose the faults,this thesis proposes a fault diagnosis method based on Latitudinal Summation Linear Discriminant Analysis(LSLDA)for identifying different faults.The method uses ECT to calculate the modulus signal to generate an initial set of samples,a transversal summation method to eliminate the non-Gaussianity of the process data,and a PCA model to reduce the redundant features of the initial data.Finally,the LSLDA method is used to solve the problem of classifying the faults.The experiments verify that the proposed method can effectively solve the fault diagnosis problem.A Cyber-Physical Systems(CPS)based fault monitoring system for MCT prototype is established,and the CPS fault monitoring task is decomposed into multiple layers according to the monitoring components and diagnosis method.Finally,the monitoring system is developed to finally realize the real-time fault monitoring of the MCT prototype.