Fault Warning And Short-term Reliability Assessment Of Wind Turbines Based On Condition Monitoring Information

Wind energy is seen as a clean and renewable energy, with huge reserves, has beenfavored by many national governments. Statistics show that wind energy of land and sea canbe developed to a total of1billion kW in China. At present, our country’s total wind powerinstalled capacity ranks first in the world, and the annual increase of installed capacity is morethan one-third of the world’s new capacity. There is no doubt that wind power has become thethird-largest source of electricity since thermal power and hydropower. China’s wind powerindustry is under sustainable development, while the fault diagnosis technology doesn’timprove that quickly, resulting in frequent failure. As a large-scale rotating machine, windturbine has a complex structure consisting of many parts, and is generally installed in remoteareas or in the ocean and the repair and maintenance cost is high. As a result, failure warningand short-term reliability assessment of wind turbines play a vital role in many aspects, suchas ensuring the stability of power generating, efficiency improvement of maintenance andrepair, optimization of the turbine’s structure, and improvement of the manufacturing process.This paper focuses on failure warning and short-term reliability assessment of windturbine to expand the study.By analyzing the impact of wind speed, power and other monitoring parameters ontemperature, this paper selected the relevant parameters to compose the relevant set ofvariables, and established the nonlinear state evaluation model of generator temperature.According to historical data of normal operation, this paper composited the process memorymatrix D, which is covering the generator running space, and selected the actual operatingdata to test the validity of the model. Verification results showed that: when the generatorworks in the normal condition, the predicted value and the actual value is in a high degree ofagreement, and residuals are small, which means the prediction model has a high precision;when the generator exists potential failure, the residual amplitude rapidly increases, andcontinuously exceeds the warning threshold, so according to the distribution of the residual, atimely measure for early warning of potential failures could be given.According to the relationship between the deviation of the parameters and systemreliability, proposed using deviation degree of parameter to characterize the decline of systemreliability, thus this establish a wind turbine reliability evaluation model of short-term.Articles using principal component analysis method to eliminate the correlation between themulti-parameter monitoring, combined entropy method to correct weights of the principal component. Using monitoring parameters of gearbox before fault were assess short-termgearbox reliability of gearbox, find its reliability continued to decline, reliability has beenreduced to a very serious level at previous time of failure, consistent with the actual situation,confirming the validity of the model. Quadratic exponential smoothing method to predictmonitoring parameters of gearbox. Predicted the reliability trend of the gearbox according tothe forecast data. Assess results is consistent, proved that parameter predicted by quadraticexponential smoothing meet requirements of reliability prediction. Combined quadraticexponential smoothing can achieve the target of predicting short-term reliability of the device.