Noise Processing Techniques For Vibration Signals Of Rotating Blade

Rotating machinery has been widely applied in the fields of China's manufacture and national defense industry.Its development would be important to our country.Blade is a key part of rotating machinery,as it should safely work to prevent the whole device failing.The vibration signal from blade should be detected real-timely as it can monitor the blade's work condition.In recent years,some technologies have been proposed to focus on monitoring the blade of rotating machinery.In our research,a vibration measurement system based on tip-timing for rotating blade is applied as it's well-developed and widely-used.We present several signal processing methods to reduce noises for blade's vibration signal based on the noises' characteristics.Also,the processed vibration signals were simulated and measured respectively,and the presented methods are tested on these signals.The main work and achievements in this thesis include:1.The noises' sources and characteristics of the measurement system are analyzed as the noises are recorded in real use.And the influence from the noises on the vibration signal measurement is discussed;2.Several noise processing techniques are studied based on the noises' characteristics,including the drift calibration,signal smoothing,digital filter and de-noising based on wavelet transform;3.The evaluation methods for the noise processing techniques in reducing noise is described.Especially,a signal-to-noise(SNR)evaluation method based on singular value decomposition has been proposed to estimate the SNR for the measured signals;4.The simulated blade vibration signals were used to test the signal processing methods.And the real vibration signals from rotating blade recorded in a well-designed filed test were also processed using these methods.The test was using a control test with dynamic strain sensor.The processing results show the improved SNRs for the signals.