Research On Mechanical Fault Diagnosis Of Wind Turbine Transmission System Based On Vibration Feature

In recent years,with the shortage of fossil energy and the increasing of environmental pollution caused by the burning of fossil energy,developing renewable clean energy vigorously is the key to solving the current problem.As a renewable energy source,wind energy plays an important role in solving the above problems,so wind power is developing rapidly.As the key component of energy conversion and transmission,the transmission system of wind turbine is influenced by its variable operating conditions and complicated structure,which leads to high failure rate.Once the parts of the transmission system fail,the spare parts replacement will be difficult,which will result in higher maintenance costs and long time machine downtime.Related research at home and abroad shows that wind turbine transmission system failure is mainly caused by mechanical reasons,so the research on machanical fault diagnosis of wind turbine transmission system has very important practical significance for enhancing the operation reliability and reducing downtime.The vibration signal generated during the running process of wind turbine transmission system contains important state information.In this paper,the mechanical state of transmission system is obtained by processing and analyzing the vibration signal of transmission system.The main research contents include three parts: signal processing,feature extraction,feature selection and condition recognition.Firstly,aiming at the non-stationary and nonlinear characteristics of the vibration signals of wind turbine transmission system,an empirical wavelet transform(EWT)algorithm is used to decompose the vibration signals into several intrinsic mode functions(IMFs).In order to improve the recognition accuracy for complicated mechanical fault of transmission system,a large number of features containing fault information are extracted from the original signal and IMFs to construct the initial feature set.Then,considering that the introduction of irrelevant and redundant data will reduce the classification accuracy and efficiency of classifier,the Gini importance of each feature is obtained in the training process of the random forest(RF)classifier,and the optimal set is then determined using sequential forward selection(SFS)according to the accuracy of the RF classifier.Finally,aiming at the overwhelming dependence on training samples of traditional multi-classification methods,a hybrid classifier composed by support vector data description(SVDD)and fuzzy c-means(FCM)clustering is constructed: firstly,a SVDD trained with normal samples is used to distinguish the normal and fault states of wind turbine transmission system.If a mechanical fault is confirmed,the fault sample and all samples with known fault types are clustered by the FCM method;then,according to the clustering results,another SVDD trained with a specific type of known fault sample is used to judge whether the fault sample belongs to a new type or not,and the final recognition results are determined.The testing results based on the measured data show that EWT can extract the fault features of transmission system accurately.The proposed hybrid classifier can not only detect the fault state accurately,but also determine whether fault samples belong to new fault types or not,which has overcome the existing shortage of over-reliance on training samples for traditional multi-class classifier.As the reliability of fault diagnosis is improved,the new method in this paper has certain engineering application value.