A Convolutional Neural Network for Impact Detection and Characterization of Complex Composite Structures

Abstract

This paper reports on a novel metamodel for impact detection, localization and characterization of complex composite structures based on Convolutional Neural Networks (CNN) and passive sensing. Methods to generate appropriate input datasets and network architectures for impact localization and characterization were proposed, investigated and optimized. The ultrasonic waves generated by external impact events and recorded by piezoelectric sensors are transferred to 2D images which are used for impact detection and characterization. The accuracy of the detection was tested on a composite fuselage panel which was shown to be over 94%. In addition, the scalability of this metamodelling technique has been investigated by training the CNN metamodels with the data from part of the stiffened panel and testing the performance on other sections with similar geometry. Impacts were detected with an accuracy of over 95%. Impact energy levels were also successfully categorized while trained at coupon level and applied to sub-components with greater complexity. These results validated the applicability of the proposed CNN-based metamodel to real-life application such as composite aircraft parts.

Method for Fault Diagnosis of Temperature-Related MEMS Inertial Sensors by Combining Hilbert–Huang Transform and Deep Learning

“…SHM monitors the failure or degeneration of components in complex systems and plays an important role in maintenance and activation inspection, researcher have used CNN to perform the complex SHM tasks. In [ 45 ], Tabian et al proposed a CNN-based meta-model to address the impact monitoring problem of complex composite structures, the method transferred the piezoelectric sensors signal to 2D images and used CNN to perform the health state classifications, this method has advantages in effective end-to-end state monitoring and well transferability to complex structures. In [ 46 ], Oliveira et al proposed a novel electromechanical impedance SHM solution by combining electromechanical impedance piezoelectricity (EMI-PZT), the high accuracy of proposed method is guaranteed by CNN-based feature extraction which include several banks of filters.…”

“…In FD applications, because raw data is often sampled in one-dimensional (1-D) format, researchers have turned to feature extraction operations that construct 2-D features for addressing FD problems using CNNs, such as sliding window [ 38 , 39 ], short time Fourier transform (STFT) [ 40 ], discrete wavelet transform (DWT) [ 41 , 42 ], and Hilbert–Huang transform (HHT) [ 43 , 44 ]. Structure health monitoring (SHM) is becoming a research hotspot in which CNN is applied and several methods have been proposed in the field of SHM combining CNN to solve mechanical system SHM problems [ 45 , 46 , 47 ].…”

A Multilevel Isolation Forrest and Convolutional Neural Network Algorithm for Impact Characterization on Composite Structures

“…CNN’s are built on multi-class classifications where each image is associated to label within a class, from which the network trains its weights. Many parameters can be represented by a class, including: a point of impact, impact localization, an energy level or any other [ 27 ].…”

“…The method in this paper utilizes a CNN as the second step for feature extraction after IF to find anomalies from the impact data. From the outputs of PZT sensors, the proposed model takes the raw data as 2D images, for different impact locations and energy levels, therefore outputs are given as the class with the highest probability [ 27 ]. Many implementations of the CNN exist, but they all include three layers: convolution, pooling and fully connected layers.…”

“…CNN’s have been used extensively in Image Processing for classification of different image models. The approach in [ 27 ] has shown that CNN’s can be sued for impact detection, localization and characterization. The data needed for training the model can be collected via Piezo-Electric sensors mounted on aircraft body (either the wings or the fuselage).…”