Abstract: In propeller synchrophasing noise reduction， the propeller noise model is identified from the measured data under certain flight conditions firstly， and then the optimal synchrophase angle is calculated based on the noise model， finally the optimal synchrophase angle will be used for propeller synchrophasing control. In the process of noise model identification， due to the influence of flight speed， height variation and air flow perturbation， the measured data often fluctuate greatly， which affects the accuracy of the identification model and the optimal synchrophase angle. In this paper， a measured data selection method based on wavelet filtering and three parameter sinusoidal fitting method is proposed to improve the accuracy of noise model identification. The noise Blade-Pass Frequency (BPF) signal is extracted through the wavelet filtering algorithm， from which the three parameter sinusoidal fitting algorithm is adopted to select the minimum fluctuation data for noise model identification. This method can effectively reduce the impact of measured data fluctuations and improve the accuracy of identification model. Experimental results has shown that the noise model identified by the minimum fluctuation data can achieve higher precision compared with the traditional noise model identified by certain data， and the error between the sound pressure level (SPL) predicted by the noise model and the actual measured SPL is less than 1dB， the error between the optimal phase angle predicted by the noise model and the actual optimal phase angle is less than 5°. Moreover， the propeller synchrophasing with the optimal phase angle can achieve 19.5dB noise reduction at the experimental location.

Key words: Propeller synchrophasing；Noise reduction；Model identification；Data fluctuation；Modeling accuracy

摘要： 相同步降噪一般先通过一定飞行条件下的实测数据辨识出螺旋桨噪声模型，然后基于噪声模型计算出该条件下的最优相角，再将最优相角用于相同步降噪。在噪声模型辨识的过程中，受飞行速度、高度和气流变化等的影响，实测数据经常会发生较大的波动，从而影响辨识模型和最优相角的准确性。提出一种基于小波滤波和三参数正弦拟合法的最小数据波动的噪声数据选取方法，提高噪声模型的辨识精度，该方法通过小波滤波算法从噪声信号中提取出螺旋桨的叶尖通过频率信号，采用三参数正弦拟合算法合理地选择出波动最小的数据用于噪声模型辨识，从而有效地回避较大波动数据，提高辨识模型的精度。试验结果表明相较于传统使用固定数据辨识所得的噪声模型，使用最小波动数据辨识所得噪声模型能够获得更高的精度，且噪声模型预测的声压级和实际测量的声压级误差小于1dB，模型预测的最优相角与实际最优相角的误差小于5°，最优相角在试验位置点能够实现高达19.5dB的降噪效果。

关键词: 螺旋桨相同步；降噪；模型辨识；数据波动；建模精度