Abstract: A new kind of health condition forecast method for the aircraft，based on the wavelet packet transform and adaptive Multi-Variable Grey Forecast Model（MVGFM） is presented in this paper to realize the veracious forecast to the health status of aircraft.The advanced Acoustic Emission（AE） technique is used to monitor the aircraft stabilizer health condition and get the AE information.The original AE signals are decomposed with the db4 wavelet packet，and the Maximum of Energy（ME），Maximum of Variance（MV），Maximum of Norm（MN） of the third layer wavelet packet decomposition structure are respectively extracted to form eigenvectors.Then the adaptive MVGFM（1，n，β） is established by the eigenvectors.The parameter β will be rectified by the errors between the forecast values and the actual ones.So the forecast precision can be adaptively improved.Experiments show that the MVGFM（1，n，β） can forecast the aircraft stabilizer fatigue crack more accurately than the GM（1，1）.And the validity of the MVGFM（1，n，β） is validated.

Key words: aircraft, daptive, avelet packet, ulti-variable grey forecast model

摘要： 针对飞行器健康状况难以准确预测的问题，提出了一种基于小波包变换与自适应多变量灰色预测模型对飞行器健康状况进行预测的新方法。采用先进的声发射技术监测飞行器关键部件的健康状态，运用小波包对由声发射监测系统募集到的飞行器关键部件原始声发射信号进行三级小波包分解，分别提取其第三级小波包分解中八个频段分解系数的能量最大值、方差最大值和范数最大值作为特征向量，并以此构建三变量MVGFM（1，n，β）模型。运用该模型对飞行器关键部件的健康状态进行预测研究，并通过该模型预测值与特征真实值之间的相对偏差来修正模型中参数β，以提高模型的下一步预测精度。实验结果表明，提出的自适应MVGFM方法可以动态实现对飞行器关键部件裂纹故障的准确预测，其预测准确度明显高于GM（1，1）模型，从而证实了该方法的有效性。

关键词: 飞行器, 适应, 波包, 变量灰色预测模型