Time series analysis and sparse sensor network-based impact monitoring for aircraft complex structures

Abstract

An impact monitoring method based on time series analysis and a sparse sensor network is proposed to identify the location of the impact event that occurs on aircraft complex structures and estimate the impact energy. Generally, a high-density sensor network is required for high localization accuracy of the impact event, but with the trade-off of high deployment costs. A sparse piezoelectric sensor network together with a coarse and fine two-step localization method is designed to balance accuracy and cost. The stress wave signals caused by impact are measured by the sensor network and regarded as time series. The impact localization problem is converted into a classification problem of time series, and the strategy of combining coarse localization (impact zone identification) and fine localization (impact localization within a zone) is used to achieve accurate identification of the impact locations. The zone where the impact event is located is first identified by the coarse localization algorithm based on K-nearest neighbor and dynamic time warping (DTW) and one-dimensional convolutional neural networks, respectively. Furthermore, the high accuracy identification of the impact location within the zone is achieved by the fine localization algorithm based on DTW and centroid localization. In terms of the impact energy estimation, the zone correction compensation algorithm is given to sufficiently reduce the estimation error of the impact energy. The low-velocity impact tests are performed on the composite stiffened panel and the aluminum alloy aircraft fuselage structure to verify the effectiveness of the proposed methods.