基于声发射信号EMD-WPD特征融合的航天器在轨泄漏辨识方法

长期运行在空间环境中的航天器可能由于撞击、振动、老化等因素而发生气体泄漏，在轨泄漏辨识对航天器安全保障具有重要意义。提出了一种基于声发射信号经验模态分解(empirical mode decomposition,EMD)和小波包分解(wavelet packet decomposition,WPD)特征融合的航天器泄漏辨识方法，首先将声发射信号分别通过EMD和WPD分解成为不同频率范围内的子带信号，考虑能量特征误差与不稳定性，提取信号无量纲因子和频率特征参数并应用Relief F算法选取特征。最后，构建支持向量机(support vector machines,SVM)机器学习数据库，训练泄漏分类模型并利用测试集交叉验证模型分类精度。结果表明，EMD和WPD分解特征并行融合分类模型可显著提高辨识精度，最高可达96.9%,且输入特征数量少，是一种具有应用前景的航天器在轨气体泄漏辨识方法。

A recognition method of spacecraft leakage based on EMD-WPD feature fusion of acoustic emission signal

Spacecrafts that have been operating in the space environment for a long time may experience gas leak due to collision, vibration, aging and other factors. Recognition of leakage in-orbit is of great significance to the safety of aerospace. A recognition method of acoustic emission signals was proposed based on the parallel feature fusion of the empirical mode decomposition(EMD) and the wavelet packet decomposition(WPD). Firstly, the acoustic emission signal was decomposed into sub-band signals in different frequency ranges through EMD and WPD, respectively. Taking into account the energy characteristic error and instability, the signal dimensionless factors and frequency characteristic parameters were extracted and the Relief F algorithm was applied to select the features. Finally, support vector machine(SVM) machine learning database was constract to train and test the leakage classification model. The results show that the proposed parallel fusion classification model can significantly improve the identification accuracy up to 96.9%, while the number of input features is small. It is a promising method for identifying spacecraft gas leakage in vacuum environment.