Fault Warning Method Of Wtnd Turbine Gearbox

With the influence of both energy crisis and environmental deterioration, the world energy supply strategy has greatly changed. Wind energy, as one of the renewables, plays an important role in the energy strategies of many countries. Wind turbines are the main equipment to tansform wind energy to electronic energy, whose installed capacity are being increased year by year. Meanwhile our country has seen more rapid development of wind energy. Correspondingly a series of issues including frequent failures and expensive maintenance costs have emerged. The gearbox, as a critical component of wind turbines, restricts the operating status of wind turbines. According to the statistics, the maintenance cost caused by gearbox faults accounts for over 40% of the total cost. Taking wind turbine gearbox as the research object, with focus on vibration monitoring analysis, the research work is carried out from both theoretical methods and system design aspects. And condition monitoring and fault warning model and system will be proposed and built based on echo state network approach, with the aim to improve fault detection accuracy and reliability and reduce false alarm rate and missing alarm rate. The research contents are mainly included as follows:Firstly, the machine condition monitoring approach based on echo state network is studied. The parameter selection and training and learning problems on echo state network are investigated Furthermore, considering the predictable characteristics of vibration feature parameters, the echo state network based monitoring model is constructed to realize the reliable prediction of the condition parameter and early fault detection. The proposed model is verified by Lorenz simulation dataset and bearing vibration dataset from the accelerated test rig.Secondly, the dynamic threshold monitoring method for wind turbine gearboxes is proposed based on echo state network. Monitoring principle and function of SCADA system installed in wind turbines is analyzed in details. To overcome the shortcomings, such as weak warning function and single alarm mode, of existingSCADA systems, the dynamic monitoring model based on echo state network and sliding window methods is built to achieve the reliable monitoring of gearbox operating status.. The historical data from SCADA system are used to validate the proposed model.(3) The fault warning system of wind turbine gearbox is designed. The system is based on Lab VIEW platform and integrates the proposed fault monitoring and detection models, which can achieve the following functions including data collection, data preprocessing, feature extraction and fault early warning and diagnosis. Finally the experiments on the wind turbine test rig are performed.and verify the feasibility and effectiveness of the designed system.