Search

Structural Health Monitoring (SHM) and Natural Fibres Composites (NFC) have livedsince early 2000’s important progress due to the needs of reducing material consumption andweight in operating structures while limiting the environmental footprint and increasing thesafety of such structures. Acoustic Emission (AE) is one of the numerous solutions that appearsto be suitable to predict service life and assess damage in composite structures. NFCs have shownpromising properties for structural applications but their complex behaviour and the limitedknowledge on their long-term durability still hinder their full development. In that sense, AEbasedSHM could allow a more important breakthrough of bio-based composites by monitoringdamage evolution and assessing their remaining life in service conditions. The new generationof AE-sensors offers also opportunities of embedment thanks to their smaller size, leading tobetter protection of the sensors and increased sensitivity. We propose in this work to study theeffect of the embedment of millimetric-size dummy sensors in flax/epoxy laminates on themonotonic tensile behaviour and damage kinetic up to failure through AE methods. Resultshighlight a limited effect of the presence of dummy sensors on the mechanical behaviour andproperties. AE results coupled with X-ray tomography pictures however highlight a change inlocation of damages that are more likely to append in the insert vicinity, and in the onset of themost severe damages, which appears at lower global deformation levels.

International audience; The paper deals with a capacitive micromachined ultrasonic transducer (CMUT)-basedsensor dedicated to the detection of acoustic emissions from damaged structures. This work aims to explore different ways to improve the signal-to-noise ratio and the sensitivity of such sensors focusing on the design and packaging of the sensor, electrical connections, signal processing, coupling conditions, design of the elementary cells and operating conditions. In the first part, the CMUT-R100 sensor prototype is presented and electromechanically characterized. It is mainly composed of a CMUT-chip manufactured using the MUMPS process, including 40 circular 100 µm radius cells and covering a frequency band from 310 kHz to 420 kHz, and work on the packaging, electricalconnections and signal processing allowed the signal-to-noise ratio to be increased from 17 dB to 37 dB. In the second part, the sensitivity of the sensor is studied by considering two contributions: the acoustic-mechanical one is dependent on the coupling conditions of the layered sensor structure and the mechanical-electrical one is dependent on the conversion of the mechanical vibration to electrical charges. The acoustic-mechanical sensitivity is experimentally and numerically addressed highlighting the care to be taken in implementation of the silicon chip in the brass housing. Insertion losses of about 50% are experimentally observed on an acoustic test between unpackaged and packaged silicon chip configurations. The mechanical-electrical sensitivity is analytically describedleading to a closed-form amplitude of the detected signal under dynamic excitation. Thus, the influence of geometrical parameters, material properties and operating conditions on sensitivity enhancement is clearly established: such as smaller electrostatic air gap, and larger thickness, Young’s modulus and DC bias voltage.

National audience; La surveillance des structures est un domaine en plein essor faisant appel à certaines technologies de capteurs qui génèrent des données dont les spécificités nécessitent de développer des algorithmes d’analyse de données dédiées. Dans cette courte communication, nous nous intéressons à la technique de l’émission acoustique dont les capteurs fournissent des données temporelles massives, non régulièrement espacées, en temps quasicontinu et sans a priori sur le lien entre les signaux d’émissions acoustiques et l’endommagement dans les matériaux. Pour cela, après une description de la technique, de l’état de l’art et de la problématique, nous proposonsune esquisse d’un modèle de mélange permettant de tenir compte de l’ordre des observations en temps continu avec possibilité de déterminer et de contraindre les instants d’apparition des classes en lien avec les endommagements.

International audience; In this work, three carbon/glass hybrid composite tubes are instrumented with eight piezoelectric wafer active sensors (PWAS), used as passive receivers of acoustic emission (AE) signals. A three point bending quasi-static loading is conducted, either in a single cycle until failure or incrementally through multiple loading/unloading cycles. In the first instance, AE signal features such as maximum amplitude, peak frequency, signal duration, and energy are used to distinguish between different damage mechanisms such as matrix cracks, delamination, and fibre breakage. The group velocity of the longitudinal modes – L(0,1) and L(0,2) – is obtained experimentally in a pitch-catch configuration between the PWAS network using the time of flight (ToF). The time of arrival (ToA) method is then used to calculate damage source locations from the received AE signals. To involve more signal features for data classification, an unsupervised clustering algorithm is applied to the datasets. Optimisation of the number of clusters is completed by maximising the robustness and minimising the uncertainty of the final result. It is found that the temporal evolution of clusters indicate the ability to distinguish between the initiation and growth of damage, as well as identifying non-damage related signals caused by extraneous noise or related to the test set-up.

International audience; Une modification des méecanismes d’inférence et d’apprentissage dans les modèles de Markov cachés autorégressifs est proposée pour prendre en compte des connaissances a priori. Ces modèles sont particulièrement intéressants pour représenter statistiquement des séries temporelles. Le fait de pouvoir ajouter des connaissances a priori permet de l’utiliser dans de nombreuses applications. Deux applications sont présentées : une première sur le pronostic avec l’estimation du temps restant avant défaillance de turboréacteurs ; et une seconde sur l’analyse d’ondes élastiques transitoires hautes fréquences rencontrées dans les structures mécaniques.

International audience; The aim of this work is to determine the first failure mode of a tubular composite/metal joint, when tested in tension. PWAS are bonded to both the composite and metal parts of the joint for detection of acoustic emissions (AE) during the tensile test. The generation of AE hits is correlated with strain as measured by surface mounted strain gauges and digital image correlation (DIC) during the test. AE data analysis is completed using an unsupervised clustering approach applied to the AE streaming data in order to identify the damage mechanisms. The approach makes use of the Gustafson-Kessel clustering algorithm to identify the clusters with arbitrary shape in the feature space. In the proposed method, different subsets of features are considered, in place of a unique subset as in standard approaches. Each subset provides one clustering result (a partition). An unsupervised information fusion process is then performed to get a representative partition, taking multiple partitions into account. The approach makes use of bootstrapped ensembles to select the number of clusters which allows to both maximiserobustness of the results according to the clustering parameterisation and evaluate the uncertainty on pattern recognition results. Subsets of features are optimised to emphasise the chronology about the onsets of AE sources and their evolution. Application of this pattern recognition chain on the AE streaming provides insights on the damage process involved in composite and metal parts of the joint.

International audience; This paper describes an Autoregressive Partially-hidden Markov model (ARPHMM) for fault detection and prognostics of equipments based on sensors’ data. It is a particular dynamic Bayesian network that allows to represent the dynamics of a system by means of a Hidden Markov Model (HMM) and an autoregressive (AR) process. The Markov chain assumes that the system is switching back and forth between internal states while the AR process ensures a temporal coherence on sensor measurements. A sound learning procedure of standard ARHMM based on maximum likelihood allows to iteratively estimate all parameters simultaneously. This paper suggests a modification of the learning procedure considering that one may have prior knowledge about the structure which becomes partially hidden. The integration of the prior is based on the Theory of Weighted Distributions which is compatible with the Expectation-Maximization algorithm in the sense that the convergence properties are still satisfied. We show how to apply this model to estimate the remaining useful life based on health indicators. The autoregressive parameters can indeed be used for prediction while the latent structure can be used to get information about the degradation level. The interest of the proposed method for prognostics and health assessment is demonstrated on CMAPSS datasets.

International audience; Due to the difficulties encountered to predict the long-term behaviour of composite structures in operating conditions, a real-time monitoring of their integrity is required in order to anticipate catastrophic failures. Although the early detection of crack initiation and propagation is beyond the potential of most non-destructive techniques (NDT), acoustic emission (AE) is one of a limited number of methods that possess the capacity for continuously detecting the occurrence of damage in large composite components or structures. Even if promising and successfully exploited in several industrial fields using commercial systems, AE has not provided at this time an effective NDT tool for composite industry, in particular for mobile structures and in-service applications. For such applications, the main difficulties are related to the real-time processing of a huge amount of complex AE time-series originating from multiple distributed sensors. One major problem is the discrimination of AE signals generated by the different damage modes from other external AE sources such as electromagnetic and mechanical noises (rubbing and friction) which are mainly generated by the surrounding environment. Another important problem is the processing of continuous and complex AE signals resulting from high AE rates, from the superimposition of transients emitted from different sources and from the distortion induced by damage accumulation. In this context, we have developed a method able to objectively discriminate with robustness AE signals generated by a specific damage mode from other AE sources in carbon fibre reinforced composite materials submitted to complex loading (multiaxial fatigue). This method includes wavelet transform-based signal processing and unsupervised multivariate pattern recognition of massive AE streaming. Uncertainties on the estimation of clusters are quantified by fusing multiple clusterings. We demonstrate on real cases that the proposed method is able to efficiently process massive AE data, as encountered in operating conditions, and take into account the distortion of the AE signals as well as the evolution of the clusters shape induced by the wave attenuation, anisotropy and damage accumulation in composite materials submitted to cyclic loading.

Safety is one of the most important considerations when buying a new car. While active safety aims at avoiding accidents, passive safety systems such as airbags and seat belts protect the occupant in case of an accident. In addition to legal regulations, organizations like Euro NCAP provide consumers with an independent assessment of the safety performance of cars and drive the development of safety systems in automobile industry. Those ratings are mainly based on injury assessment reference values derived from physical parameters measured in dummies during a car crash test. The components and sub-systems of a safety system are designed to achieve the required restraint performance. Sled tests and other types of tests are then carried out by car makers and their suppliers to confirm the protection level of the safety system. A Knowledge Discovery in Databases (KDD) process is proposed in order to minimize the number of tests. The KDD process is based on the data emerging from sled tests according to Euro NCAP specifications. About 30 parameters of the passive safety systems from different data sources (crash data, dummy protocol) are first analysed together with experts opinions. A procedure is proposed to manage missing data and validated on real data sets. Finally, a procedure is developed to estimate a set of rough initial parameters of the passive system before testing aiming at reducing the number of tests.