Artificial neural network ensembles for fatigue damage detection in aircraft

Abstract

Neural networks are commonly recognized tools for the classification of multidimensional data obtained in structural health monitoring (SHM) systems. Their configuration for a given scenario is, however, a challenging task, which limits the possibilities of their practical applications. In this article the authors propose using the neural network ensemble approach for the classification of SHM data generated by guided wave sensor networks. The overproduce and choose strategy is used for designing ensembles containing different types and sizes of neural networks. The proposed method allows for a significant increase of the state assessment reliability, which is illustrated by the results obtained from the practical industrial case of a full-scale aircraft test. The method is verified in the process of detecting fatigue cracks propagating in the aircraft load-carrying structure. The long-term experiments are performed in variable environmental conditions with a net of structure-embedded piezoelectric sensors.

Comparison of Novelty Detection Methods for Detection of Various Rotary Machinery Faults

“…Therefore, each network, trained on the same training set, gives different normal regions, as presented in Figure 3b. Simultaneous training of several networks and constructing an ensemble, in which their responses are averaged together, producing a more reliable representation of a normal region in feature space [34][35][36].…”

Identification of the stick and slip motion between contact surfaces using artificial neural networks

“…One of the many methods to solve this issue require training of several ANNs, then choice of few that are best in validation data assessment and finally, in operational phase, average these responses to obtain an ensemble output. This ensemble approach usually boosts reliability and accuracy of the ANN-based diagnose [18] and was successfully used in various mechanical systems research scenarios [19][20][21].…”

“…Such approach, known under ensemble name ,is used in this paper as well. A simple overproduce and choose approach is used: A fraction of ANNs that scored best during training are stored as ensemble members [21,38]. In Fig.…”

Damage detection in structural systems utilizing artificial neural networks and proper orthogonal decomposition

“…Recently, Jin et al used an extended Kalman filter for estimating weights of a regression neural network for damage detection of a highway bridge under severe temperature changes . Dworakowski et al proposed a classification neural network for fatigue damage detection in aircraft . Gu et al proposed a framework based on ANN for removing false positive SHM alarms stemming from temperature variations .…”

“…41 Dworakowski et al proposed a classification neural network for fatigue damage detection in aircraft. 42 Gu et al proposed a framework based on ANN for removing false positive SHM alarms stemming from temperature variations. 43 Sbarufatti proposed a framework for optimizing ANN hyperparameters based on analysis of variance and used the optimized ANN for fatigue damage identification.…”