Study On Condition Monitoring And Fault Diagnosis Of Doubly-fed Wind Generator

Affected by alternating loads and natural factors,wind turbines are prone to abnormal operation,and long-term operation with faults will cause major losses such as damage to various components and even shutdown.If the maintenance plan is arranged in time to avoid the risk of failure and prevent the further development of the failure,the operation and maintenance cost of the wind power system can be greatly reduced.When a fault occurs,the system operating parameters will change.By monitoring and analyzing these parameters and designing a fault diagnosis model,the early fault monitoring capability of the fan can be greatly improved.This paper mainly designs the fault detection,fault reconstruction and fault identification algorithms for common faults that may occur in the operation of wind turbines.The research content is divided into the following three aspects:1.A fault detection method for doubly-fed induction generators(DFIG)based on PSO-SMO is proposed.The traditional sliding mode observer can achieve effective fault detection by reconstructing the doubly-fed induction generators model and comparing it with the measurable state quantity.However,unreasonable sliding mode observer parameters will greatly reduce the accuracy of fault detection and even cause false alarms.Aiming at the difficulty of selecting sliding mode parameters,this paper proposes to combine particle swarm optimization(PSO)algorithm with sliding mode observer for fault detection of DFIG.This method can obtain extremely high observation accuracy while minimizing chattering in the observer.First,SMO is designed based on the mathematical model of the DFIG.Then,the PSO algorithm is used to find the optimal sliding mode observer gain.Finally,the normal operating conditions,grid terminal voltage sag fault and rotor current sensor fault are set,and the effectiveness of the sliding mode observer for rotor current monitoring under complex conditions is verified by simulation experiments.2.A sensor fault reconstruction method based on sliding mode observer is proposed.The method can estimate sensor faults and compensate the measured values accordingly,so that the doubly-fed induction motor can maintain stable operation.First,sensor faults are considered in the mathematical model of DFIG,and the augmented state space equations of the system are established by constructing virtual states to convert sensor faults into actuator faults.Then a sliding mode observer is established for the augmented system to reconstruct the fault value.According to the reconstructed fault signal value,DFIG fault detection is realized.Finally,taking the rotor current sensor as the observation object,the deviation fault and drift fault of the sensor are respectively set.The sensor fault reconstruction method is verified by simulation experiments.3.A fault diagnosis method based on virtual extension and spherical mapping model is proposed to solve the problem of limited fault data sets in the actual operation of wind turbines.Firstly,Hermite interpolation is performed on the discrete wind power data samples to obtain the interpolation curve about the sample characteristics,and uses the synchronous sampling method to construct virtual samples for the interpolation curve.Then,the features of the virtual sample are mapped to a three-dimensional space.Define the spherical data model and perform spherical fitting in a three-dimensional coordinate system.Finally,feature extraction is performed on the fitted spherical surface for training and testing ELM.The distribution law of fault data in the spherical model is summarized.Using the data generated based on the Bootstrap method as a control group,comparative experiments were carried out in BP,PNN,GRNN,and SVM,which verified the effectiveness of the proposed method.