Spacecraft Leakage Detection Using Acoustic Emissions Based on Empirical Mode Decomposition and Support Vector Machine

摘要

Spacecraft operating in the space environment for a long time may leak due to collision, vibration, aging and other factors, which would pose a threat to the aerospace safety. The realization of spac ecraft leakage monitoring and identification is of great significance. This paper presents a leak detection algorithm by using acoustic signals based on Empirical Mode Decomposition (EMD) and Support Vector Machine (SVM). In order to eliminating effect of measurement error of signal energy of acoustic emission sensor, the non-dimensional factors and frequency statistical features are used. First, the leakage signal is decomposed by EMD. Then the dominant component of the intrinsic mode function (IMF) is selected to extract the dimensionless factors and frequency characteristic parameters. The ReliefF algorithm is applied to select the most discriminative features and the feature set is selected as the input for SVM classification model based on ReliefF algorithm results. Finally, the results show that the method used all the extracted features to determine the leakage state and the non-leakage state with an accuracy of 90.6%, the accuracy of using the first six most distinguishing features is 93.8%, and the classification accuracy of using the first eighteen features can reach 98%.