| Wind blades are made of fiber glass composites and weight-reducing balsa wood,PVC foam panels.They are also the main bearing structures in wind turbines.Wind blades are subjected to wind alternating and fatigue load for a long time,The area rich in resin is prone to damage,suffer from thunderstorms,strong winds,sandstorms and other erosion,the surface will inevitably appear holes,trachoma and other defects.If not found in time the expansion will cause greater damage light reduce service life to shut-down.Because of its complex structure and large size,it is difficult to carry out the general detection method in service.A nondestructive testing method is urgently needed to investigate the damage mechanism of the blade and to provide protection for service blades.In this paper,Acoustic Emission technology is used to monitor the damage of wind blades in real-time,and the blade damage pattern is identified by advanced data analysis methods such as cluster analysis,spectrum analysis and wavelet packet transformation.Subsequently using the monitoring data put forward a damage monitoring scheme for blades with large scale to provide the damage was warned immediately.The specific research contents are as follows:First,the single component epoxy resin,fiber bundle specimen,multi-component glass fiber specimen with different paving direction(0°,90°,45°)and small blade preformed crack specimen were made,and the damage AE was detected,receive the frequency band range of resin fracture and fiber fracture AE signal is 0 MHz-0.06 MHz,0.06 MHz-0.12 MHz respectively.Subsequently the k-means cluster analysis of 45° laminates and mini-fan samples was carried out to find out the number of damage types and the best characterization parameters.According spectrum analysis,we found two new frequency ranges of 0.25 MHz-0.32 MHz and 0.12 MHz-0.19 MHz in addition to the first two damage modes as fiber fracture and fiber extraction,respectively.Secondly,we used wavelet packet analysis to identify wind turbine blade damage pattern,measured disassembly scale according to the results of spectrum analysis,calculated time-frequency characteristics of different damage patterns by wavelet packet transform,and used wavelet packet spectral coefficient to identify four damage patterns.It is concluded that different layering modes and small blade damage evolution process are obtained.It is found that 90° samples are mainly matrix fracture,0° samples are mainly fiber fracture,and 45° delamination propagation occurs in the damage process of samples.The previous three kinds of damage appeared in the first three times of small leaf injury,and the fourth one appeared delamination expansion.Third,the initial stage of no damage fatigue,the middle stage of minor damage fatigue and the later stage of obvious damage fatigue are the three stages of large blades fatigue monitored for a long time.The scheme of monitoring was making 10-min test with 20-min interval.EMD-Wavelet threshold is used to reduce the noise of AE signal.Concluded that the average Amplitude and Count in early and middle fatigue does little different,and later period,the average Amplitude and Count increases obviously.After analyzing the monitoring date,it is found that there are obvious differences in Amplitude and Count of before and after fatigue.Calculation of wavelet packet energy spectrum coefficients for detection date,It is found that there was no damage signal in the early stage,matrix fiber fracture and a small amount of fiber extraction signal in the middle stage,delamination expansion signal in the later stage.The damage monitoring system based on amplitude counting and wavelet packet energy spectrum is designed by using the obtained data to warn the damage. |
PDF Full Text Request