Maintenance Management in Wind Turbines by Monitoring the Bearing Temperature

Abstract

Wind turbines are increasing in number, size and market share. It is determined whether they are efficient through operating and maintenance costs. Therefore, one of the main objectives of the wind turbines is to increase the service life of the components by applying different methodologies for fault detection. The gearbox is a critical component since it causes the most downtime and failure rate of the wind turbines. The Supervisory Control and Data Acquisition system offers the measurement of several variables, and by a correct analysis it is possible to detect the faults before they occur. This paper analyses the temperature curve of bearing versus wind speed as significant variables of a gearbox failure for fault detection.

False Alarms Analysis of Wind Turbine Bearing System

“…The variables related to alarms, the state and output power, are analyzed together considering the alarm manual, leading to the correct condition of the WT. Figure 4 shows that the measurements before the alarm activation appear within the confidence interval employing only the polynomial 7 used in reference [42]. However, in Figure 5, where the approach is applied, those measurements are classified as outliers, allowing the identification of critical alarm periods outside the confidence interval.…”

“…For this real case, only average values are used by the SCADA data for each 10 min time interval of gearbox bearing temperature, wind speed, and power output. The data acquired are fitted to a polynomial of degree 7, according to a previous study [42]. The polynomial of degree 9 has a quadratic mean error (RMSE) of 4.808, and the polynomial of degree 7 has an RMSE of 4.809.…”