Prognostics and health management techniques have drawn widespread attention due to their ability to facilitate maintenance activities based on need. On-line prognosis of fatigue crack propagation can offer information for optimizing operation and maintenance strategies in real-time. This paper proposes a Lamb wave-particle filter (LW-PF)-based method for on-line prognosis of fatigue crack propagation which takes advantages of the possibility of on-line monitoring to evaluate the actual crack length and uses a particle filter to deal with the crack evolution and monitoring uncertainties. The piezoelectric transducers (PZTs)-based active Lamb wave method is adopted for on-line crack monitoring. The state space model relating to crack propagation is established by the data-driven and finite element methods. Fatigue experiments performed on hole-edge crack specimens have validated the advantages of the proposed method.

Accurate prognosis of fatigue crack growth is of great importance to ensure structural integrity, which is a challenging task due to various uncertainties affecting crack growth. To deal with this problem, the particle filter (PF) based prognostics that incorporates on-line structural health monitoring (SHM) becomes a new trend. However, most existing studies adopt the basic PF algorithm, which needs improvements to meet the requirement for on-line prognosis. It refers to the choice of the importance density and the resampling strategy, as well as the definition of the measurement equation that correlates SHM data to crack states. Till now, no literature addresses this topic in-depth. Aiming at on-line crack growth prognosis, this paper combines four improved PFs with the guided wave based SHM. The study is carried out under two cases respectively, which involve whether or not the measurement equation is accurately trained based on fatigue test data of a kind of aircraft attachment lug. Not only prognostic accuracy and consistency, but also effects of the particle number on the performance and computational cost are analyzed. The result shows advantages and disadvantages of each improved PF for on-line crack growth prognosis, giving instructions to choose appropriate PFs for different application scenarios.

Fatigue cracks are one of the common failure types of key aircraft components, and they are the focus of prognostics and health management (PHM) systems. Monitoring and prediction of fatigue cracks show great application potential and economic benefit in shortening aircraft downtime, prolonging service life, and enhancing maintenance. However, the fatigue crack growth process is a non-linear non-Gaussian dynamic stochastic process, which involves a variety of uncertainties. Actual crack initiation and growth sometimes deviate from the results of fracture mechanics analysis. The Lamb wave-particle filter (LW-PF) fatigue-crack-life prediction based on piezoelectric transducer (PZT) sensors has the advantages of simple modeling and on-line prediction, making it suitable for engineering applications. Although the resampling algorithm of the standard particle filter (PF) can solve the degradation problem, the discretization error still exists. To alleviate the accuracy decrease caused by the discretization error, a Lamb wave-minimum sampling variance particle filter (LW-MSVPF)-based fatigue crack life prediction method is proposed and validated by fatigue test of the attachment lug in this paper. Sampling variance (SV) is used as a quantitative index to analyze the difference of particle distribution before and after resampling. Compared with the LW-PF method, LW-MSVPF can increase the prediction accuracy with the same computational cost. By using the minimum sampling variance (MSV) resampling method, the original particle distribution is retained to a maximum degree, and the discretization error is significantly reduced. Furthermore, LW-MSVPF maintains the characteristic of dimensional freedom, which means a broader application in on-line prognosis for more complex structures.

Based on the theory of modal acoustic emission (AE), when the convolutional neural network (CNN) is used to identify rotor rub-impact faults, the training data has a small sample size, and the AE sound segment belongs to a single channel signal with less pixel-level information and strong local correlation. Due to the convolutional pooling operations of CNN, coarse-grained and edge information are lost, and the top-level information dimension in CNN network is low, which can easily lead to overfitting. To solve the above problems, we first propose the use of sound spectrograms and their differential features to construct multi-channel image input features suitable for CNN and fully exploit the intrinsic characteristics of the sound spectra. Then, the traditional CNN network structure is improved, and the outputs of all convolutional layers are connected as one layer constitutes a fused feature that contains information at each layer, and is input into the network’s fully connected layer for classification and identification. Experiments indicate that the improved CNN recognition algorithm has significantly improved recognition rate compared with CNN and dynamical neural network (DNN) algorithms.

This paper presents empirical path loss models for the environment of airplane wings. Two wireless prototypes of monitoring-of-airplane-wing (MOAW) in 433, 868, 780, 915 and 2400 MHz band is considered for monitoring propagation in a delta airplane wing. Path loss models for the five bands are proposed for spar-mounted, rib-mounted, front-stringer-mounted and rear-stringer-mounted wireless link scenarios inside an airplane delta wing. Moreover, the packet loss rates (PLR) of wireless prototypes at five bands inside airplane wing are investigated. The results of wireless link budget calculations inside airplane provide insights into how the MOAW devices are deployed inside airplane wing for constructing better propagation, and give advisable reference to the deployment of devices inside more types of airplane wings in future.