Nondestructive Testing Data Fusion Research Articles

Development and fusion of NDT classifiers for defect detection on underwater structures

Vessel hull inspection is a complex and labor-intensive task, motivating the utilization of data fusion from complementary nondestructive testing (NDT) techniques to boost reliability and reduce detection uncertainties. In this study, an underwater robot equipped with three NDT systems – monocular RGB camera, eddy current (ECT), and ultrasound (UT) – is employed to assess a 12.7 mm thick ASTM-A36 carbon steel plate with artificially induced defects – coating loss, crack, pitting, and generalized corrosion. A classifier is developed for each NDT system for defect classification. Mask R-CNN is used for the camera data, achieving a mean average precision of 67.7% across three defect classes. A multilayer perceptron is used to classify the ECT data into six defect types and the non-defective class, attaining an accuracy of 98.8% on the test set. For UT data, a rule-based classifier based on A-scan-derived features distinguishes four defect classes. Subsequent to spatial synchronization, these classifier outputs are fused at the decision level, where redundancies between the camera and ECT classifiers are eliminated. Their outcomes are concatenated with the UT classifier results, culminating in a comprehensive detection map capable of representing eleven distinct defect classes. The classifiers and fusion are evaluated on a novel plate with unseen artificial defects. The camera and UT classifiers excel in detecting all classes for which they were specifically trained. In contrast, ECT exhibits a more limited capability, detecting only 21% of the specified defect classes. The findings underscore the potential of NDT data fusion to enhance inspection reliability by minimizing redundancies and extending the detection range.

Non-Destructive Health Monitoring of Structural Polymer Composites: Trends and Perspectives in the Digital Era.

Non-destructive testing (NDT) techniques are usually used for the characterisation of defects arising in polymer composites during manufacturing or in-service use. However, each of these NDT techniques cannot always allow a full diagnosis of the material's or component's structural health. Thus, several techniques have to be combined in order to improve the diagnosis of the damaged state of composite structures and their evolution during the part's life span. This opinion paper proposes a critical overview of the use and applicability of these NDT techniques for the detection and characterisation of damage to structural composite materials in view of in-service performance assessment and residual durability prognosis. It also addresses some current trends of structural health monitoring (SHM) of these materials, such as sensor-actuator embedding and NDT data fusion, and draws future perspectives on how composite SHM could evolve in the digital era, taking advantage of artificial intelligence, Internet of Things and big data to implement digital twins.

Pulsed eddy current and ultrasonic data fusion applied to stress measurement

Stress measurement and its variation are key problems in the operating performance of materials. Stress can affect the material properties and the life of components. There are several destructive and nondestructive techniques that are used to measure stress. However, no single nondestructive testing (NDT) technique or method is satisfactory to fully assess stress. This paper presents an NDT data fusion method to improve stress measurement. An aluminum alloy 2024 specimen subjected to stress simulation is nondestructively inspected using pulsed eddy current and ultrasonic techniques. Following these nondestructive examinations, the information gathered from these two NDT methods has been fused using a suitable fuzzy combination operator. The results obtained with these processes are presented in this paper and their efficiency is discussed. It is shown that the fusion of NDT data with a suitable fuzzy operator can be adequate to improve the reliability of stress measurements.

NDT Data Fusion for the Enhancement of Defect Visualization in Concrete

Ground penetrating radar (GPR) and active thermography are well known non-destructive testing (NDT) methods for structural visualization and defect detection in concrete. However, for both methods, the probability of detection is strongly depth-dependent and each method suffers from an almost blind region at a specific depth. In this study we propose the use of unsupervised clustering techniques for the fusion of GPR and thermographic phase contrast data to enhance defect visualization in concrete. The evaluation was carried out on the basis of experimental data acquired on laboratory concrete test specimens, which contain inbuilt anomalies varying in shape, material and position. To achieve an optimal fusion of radar depth slices and thermographic phase contrast images along the depth axis, we derive sensitivity curves for both NDT methods and use the probability mass information to further improve the fusion results. Results show that the fuzzy c-means algorithm may contribute to an enhanced detection probability of defects below high density reinforcement. For defects with a concrete cover from 1.5 to 2 cm, the use of weighted clustering is particularly suggested. In general, complex defect types and shapes could be better resolved by using the Gustafson-Kessel algorithm or noise clustering. In addition, we demonstrate the application of the Dempster-Shafer theory to quantitatively evaluate the effectiveness of fused data on the basis of joint mass probability.

Concrete properties evaluation by statistical fusion of NDT techniques

Measurements from Non-Destructive Testing (NDT) techniques are affected in different ways by concrete properties such as porosity, complexity of the pore network, water content, strength, etc. Therefore, extracting one concrete property from one NDT measurement appears to result in uncertainties. This highlights the benefit of NDT data fusion to evaluate accurately concrete properties. In this paper, NDT measurements from GPR, electrical resistivity and ultrasonic pulse velocity were combined to predict more accurately concrete properties such as strength and water content. Two techniques of data fusion were used namely Response Surface Method (RSM) and artificial neural networks (ANNs).The results obtained show the effectiveness of the statistical modeling to predict the properties of concretes by fusion of NDT measurements. In the context of this study, the performances of the two techniques of fusion appear relevant in terms of water content and concrete strength prediction. ANN models exhibit better predictive ability than RSM ones.

Experimenting with pixel-level NDT data fusion techniques

Results from the fusion of images from multiple nondestructive testing (NDT) sources gathered during the inspection of composite material damaged by impact are presented and discussed. The fusion of multiple images was performed at pixel level using several data fusion techniques based on Bayesian analysis, wavelet theories (Daubechie wavelet and steerable pyramid transform) and multi-resolution mosaic technique. Based on assessment of the resulting fused images, it appears that the multi-resolution mosaic approach is best suited to offer an estimate of the actual damaged area.

NDT data fusion at pixel level

No single non-destructive testing (NDT) technique or method is satisfactory to fully assess the structural integrity of a material. Indeed, each method presents some limitations in terms of defect detection and characterisation. Additionally, poor signal-to-noise ratio may make signal interpretation complex or unreliable. For these reasons, the concept of NDT data fusion — based on the synergistic use of information from multiple sources in order to facilitate signal interpretation and increase defect detection and characterisation — is expanding rapidly. A carbon fibre reinforced composite material previously subject to low energy impact damage has been non-destructively inspected using eddy current and infrared thermography. Following these non-destructive examinations, the information gathered from these two NDT methods has been fused at pixel level. Several well established and recognised data fusion processes based on statistical and probabilistic algorithms were used to combine eddy current and infrared thermographic data. The results obtained with these processes are presented in this paper and their efficiency discussed. It is shown that, in certain cases, NDT data fusion at pixel level may be adequate to increase knowledge about defect location and characterisation and to reduce ambiguity.