Research On Gear Fault Warning And Diagnosis Of Wind Turbine Gearbox

General structure of the planetary transmission has the features of small size and large transmission ratio, it gradually replaces the traditional gearbox, which is widely used in various large-scale machinery and equipment, once the planetary gearbox failure occurs, it will cause a lot of accidents and economic losses, traditional early warning and fault diagnosis for gearbox transmission structure is relatively mature, but the study of planetary gearbox failure is actually less, aiming planetary gearboxes of wind turbine, the paper uses SCADA data and vibration data to study early warning and fault diagnosis.The main contents of this paper are as follows:1. First, the common faults gear were introduced, starting from the vibration mechanism gears, gear vibration build model to analyze the vibration signal characteristics under typical fault gear, while the type of wind turbine gearbox and structure are introduced, then study the gear wear fault, combined with FTA and FEMA technology to analyze the cause, effect, signs and corresponding maintenance measures to establish a complete fault knowledge base.2. Little use for the status of using the existing SCADA data to conduct warning and diagnosis of gearbox fault, this paper studys the gearbox lubricant oil, Aiming at the problem that the operational condition of large size wind turbines is complex and relying on constant lubricating oil temperature as gearbox fault early warning values is easily wrong, an online fault early warning method for wind turbine gearbox based on operational condition division was proposed.By dividing different operation conditions, this method respectively set threshold for different operation conditions according to the Gaussian Model. Take the real-time data into the corresponding operational condition to determine whether the data is abnormal, and use moving window to calculate the abnormal rate as a indicator to trigger gearbox fault early warning.3. On the basis of the parallel gear vibration,we analysis planetary gear train vibration signal analysis model to calculate the rotation frequency of the various components of the planetary gear train, respectively, analysis the planetary gear train vibration signal model when it is normal、 distributed fault or local failure,get the characteristic frequency of each failure, using MATLAB software to simulate each state planetary gear train, and analyze the spectral characteristics of each state.4. Dimensionless index is sensitive to early gear failure,can be used as an early warning of gear failure, the need for a new dimensionless index is calculated on the basis of the entire period of the sampling requirements and operating conditions of the wind turbine is unstable, the measured vibration signal is non-stationary signals, using order analysis and EMD decomposition to preprocess vibration signal, and then transformed into a smooth angle-domain signal, while the gearbox failure of different sensors measuring points provide different sensitivity of the information or complementary information, the paper uses multi-sensor vibration signal measured dimensionless parameter as warning indicators, using genetic algorithms to optimize neural network to conduct gear failure early warning is faster than general neural networks with high accuracy.5. On the basis of the planetary gear train fault feature of the second point, we propose a new gearbox fault location and severity of the online diagnostic method, first extract fault information that can accurately characterize the features of the amount, and then calculate and J-divergence and KL-divergence values for each failure mode between preset standard sample and the test sample,the use of the divergence values can characterize the failure mode and the change of the value of divergence can trace fault severity, the characteristics can determine the fault location and severity, while the lack of wind turbines for the fault sample, this method is improved by calculating the value of the divergence between the different fault characteristic of the test sample and the normal standard sample, you can effectively determine the failure mode and severe degree.6. On the basis of the study of this paper,we apply the research content of this article to the actual fault warning and diagnosis of wind turbine planetary gearbox, test and verify the validity of this method, at the same time we conduct the initial development of the wind turbine gearbox fault warning and fault diagnosis system.