In this paper, to increase the performance of the sparse reconstruction method in real complex engineering structures an adaptive dictionary learning framework is proposed which updates the dictionary matrix, to allow improved compatibility with the complex structure. This proposed framework was developed by combining analytical modeling with training data sets and learning methods. An experimental evaluation of the proposed dictionary learning framework was performed on an anisotropic composite plate with a stiffener. In this experimental evaluation, a moving magnet was used as the artificial damage to capture the training data set, and both artificial damage in several locations and real impact damage was used for detection and location of the target damage. The obtained results confirmed the concept of the proposed dictionary learning framework for the improved health monitoring of complex structures.@en

To perform active structural health monitoring (SHM), guided waves (GW) have received great interest as they can inspect large areas with a few sensors and are sensitive to barely-visible structural damages. Fiber Bragg grating (FBG) sensors offer several advantages such as small size, low weight and ability to be embedded but their use has been limited for GW sensing due to their limited sensitivity while using spectrometers. FBG sensors in the edge-filtering configuration have overcome this issue with reasonable sensitivity and there is a renewed interest in their use. It is well known that when subjected to a transverse strain, the circular cross-section of the fiber deforms into an elliptical shape generating the birefringence phenomenon. This deformation, influences the coupling mode of the light inside the FBG and hence, modifies the resulting reflectivity spectrum. This paper investigates how controlled changes in the reflectivity spectrum can be introduced using different transverse loads. The effect of the modified spectrum on the sensitivity of the FBG for GW measurements is then studied. The study also investigates the effect of the transverse strain on the coupling of the GW from the structure into the fiber.@en

In this paper, the performance of the sparse reconstruction (SR) and the delay-and-sun (DAS) methods for damage localization, were evaluated for various environmental and operational conditions, both numerically and experimentally. To assess these damage localization methods, a methodology based on the Taguchi method was used to make the experimental design, and a modified performance-index was defined to represent the quality of reconstructed images. Then, the robustness and the accuracy of each method, in a well-defined performance region relevant to in-service aerospace structures, were investigated using the Taguchi and analysis of variance methods. It was concluded that for the defined conditions, the robustness of the delay and sum method is better than the sparse reconstruction method for uncontrolled factors. However, the sparse reconstruction method is more robust to poor baseline subtraction than the delay and sum method, while the delay and sum method was more robust to factors that lead to a model mismatch. These results provide additional insight into the design of reliable accurate structural health monitoring systems. The outcomes of this work can be used in future reaserch into SHM imaging techniques.@en

To perform active structural health monitoring, guided Lamb waves for damage detection have recently gained extensive attention. Many algorithms are used for damage detection with guided waves and among them, the delay-and-sum method is the most commonly used algorithm because of its robustness and simplicity. However, delay-and-sum images tend to have poor accuracy with a large spot size and a high noise floor, especially in the presence of multiple damages. To overcome these problems, another method that is based on sparse reconstruction can be used. Although the images produced by the sparse reconstruction method are superior to the conventional delay-and-sum method, it has the challenges of the time and cost of computations in comparison with the delay-and-sum method. Also, in some cases in multi-damage detection, the sparse reconstruction method totally fails. In this article, using prior support information of the structure achieved by the delay-and-sum method, a hybrid method based on sparse reconstruction method is proposed to improve the computational performance and robustness of sparse reconstruction method in the case of multi-damage presence. The effectiveness of the proposed method in detecting damages is demonstrated experimentally and numerically on a simple aluminum plate. The technique is also shown to accurately identify and localize multi-site damages as well as single damage with low sampled signals.@en

The performance of the Acoustic Emission (AE) technique is significantly dependent on the sensors attached to the structural surface. Although conventional commercially AE sensors, like R15a and WSa sensors, have been extensively employed in monitoring many different structures, they are unavailable in restricted-assess areas. In contrast, thin PZT sensors are small, inexpensive and lightweight. These thin PZT sensors have a great potential for passive sensing to detect AE signals. However, their utility in AE monitoring is limited due to their low signal-to-noise ratio and information incompleteness because of their simple construction. This work discusses the issues and possible solutions with regards to the specific selection and application of thin PZT sensors for passive sensing. The compatibility of different thin PZT sensors and conventional bulky sensors is investigated using pencil break lead (PBL) tests. The comparison between the recorded signals is carried out in both the time domain and frequency domain for these sensors. To improve the reliability and performance of the thin PZT sensors, a methodology employing multiple thin PZT sensors of different sizes is proposed based on machine learning techniques and sensor data fusion.@en