Wind turbine blade damage detection based on blind separation of acoustic signals and fixed point iterative algorithm

Due to the weak strain signal changes generated on the surface of wind turbine blades when they are damaged, it is difficult to accurately extract damage characteristics, thereby reducing the detection accuracy of damage types. Therefore, a wind turbine blade damage detection method based on blind separation of acoustic signals and fixed point iterative algorithm was proposed. Based on the simulation model of the wind turbine, sound sensors were used to collect the acoustic signals on the surface of the wind turbine blades. Blind separation technology was introduced to demodulate the signals, independent source signals were obtained, and iteratively converge was carried out to obtain the surface strain signals of the wind turbine blades. Combined with the blade structure vibration equation and the damage frequency domain function, the vibration strain mode difference was calculated to extract the damage characteristics. Support vector machine algorithm was used to construct a blade damage detection model, and damage features were used as input and damage types were used as output to achieve detection of wind turbine blade damage. The experimental results showed that under the application of the proposed method, the misjudgment rate of wind turbine blade damage detection results was always controlled below 0.20%, and the detection accuracy was relatively high.