Detecting anomalies, such as defects in newly manufactured products or damage in long-used material structures, is a tedious task for humans. Given the current advances in the areas of artificial intelligence (AI) and computer vision, automation of visual quality control is possible and can be a reliable solution. Current methods that achieve state-of-the-art anomaly detection results often use features of different scales extracted from pre-trained convolutional neural networks (CNN). We propose employing multiple feature sets of the same scale alongside a normalizing flow model designed specifically for such sets. Such input features allow the creation of significantly smaller flow models with faster inference. We managed to achieve a decrease of up to 77.5% in the number of flow model parameters and 63.6% in inference time while still accomplishing results better than all other flow models and all but one non-flow method. Experimental evaluation on publicly available MVTec AD and MTD datasets showed a state-of-the-art level of performance of our models, thus proving that it is not necessary to use different scales to detect anomalies of different sizes. This research paves the way for real-time, efficient AI-based automation of visual inspection. Graphical abstract Anomaly detection pipeline of FRAnomaly

The safety of nuclear power plants has always been one of the most important security issues in the industry in general. Numerous standards, techniques, and tools have been developed to deal specifically with the safety of nuclear power plants – one has specialised probes, robotized systems, electronics, and software. Although seen as a mature (or slowly evolving) industry, this notion about nuclear safety is a bit misleading – the area is developing in many promising new directions. Some recent global events will speed up this development even more. On the other hand, the industry is currently going through digital transformation, which brings networking of devices, equipment, computers, and humans. This fourth industrial revolution promises speed, reliability, and efficiencies not possible up until now. In the NDE sector, new production techniques and traditional manufacturing lines are getting to be lights-out operations (near-total automation). The same is most probably going to happen with the safety inspections and quality insurance. Robotics and automation are improving worker safety and reducing human error. The well-being of inspectors working in a hazardous environment is being taken care of. Most experts agree that the digitalization of NDE offers unprecedented opportunities to the world of inspection for infrastructure safety, inspector well-being, and even product design improvements. While the community tends to agree on the value proposition of digital transformation of NDE, it also recognizes the challenges associated with such a major shift in a well-established and regulated sector. The work presented in this paper shows a part of the project that aims to develop a modular ultrasound diagnostic NDE system (consisting of exchangeable transducers, electronics, and acquisition/analysis software algorithms), for applications in hazardous environments within nuclear power plants. The paper will show how the software part of this system can reach near-total automation by implementing various deep learning algorithms as its features and, then, testing those algorithms on laboratory samples, showing encouraging results and promises of online monitoring applications. Furthermore, future general prospects of this technology are discussed, and how this technology can affect the well-being of nuclear power plant inspectors and contribute to overall plant safety.

High temperature pipe cracks are the root of a steam power failure in the EU typically every 4 years, resulting in loss of human life, serious accidents and massive financial losses. According to IAEA’s Reference Technology Database, such an event on a nuclear power plant has an average cost of €120 million, including outage costs, emergency repair costs, insurance and legal costs. Since only one growing crack is needed to cause a major failure, they have to be inspected and monitored thoroughly. Breakdowns at extreme conditions (e.g. 580°C, 400 bar) are a result of two major weld failure modes: a) creep cracks near pipe welds; b) fatigue cracks on pipe welds. Current maintenance practice is to proceed with repairs on a detected crack according to its severity. For cost reasons, cracks that are not judged as severe enough will not be repaired. Crack severity judgement is based on its probability to cause a failure and this probability is derived taking into account the crack size and operational lifetime. More variables such as operating temperature and vibrations may rarely be found in other studies. Recent data from fracture mechanics statistical studies shows this connection between the size of a crack on a nuclear power plant pipe and its probability to lead to a failure. To deal with the above problems two Structural Health Monitoring (SHM) systems have been developed and they are presented in this work. These systems are able to achieve continuous operation for an extended time period at operating temperatures of nuclear power plants. The developed systems employ novel phased array (PA) ultrasonic and ultrasound guided wave (UGW) probes able to withstand and continuously operate even up to 580 °C. The systems are designed to be permanently mounted on superheated steam pipes, at locations of known defects and to continuously monitor their size. However, this supposes that defects will have already been detected by a traditional method during an outage. The PA transducers are placed according to the Time-of- Flight Diffraction (TOFD) technique’s topology, thus creating a novel configuration, while the UGW transducers are placed on a stainless steel ring in a circular array configuration. These configurations can enable continuous tracking of cracks growth with high accuracy, enabling maintenance crews to estimate the severity directly and not through statistics.

Failures of welded main steam pipes in power station plants are related to cracks forming and propagating in the fine-grained zone, so called heat-affected zone (HAZ). Experience shows that Type IV cracks are indeed some of the most common causes of creep failure for low-alloy ferritic steel pipes and headers in power stations. Typically, crack damage will be characterised by high strain rates in the HAZ and local grain boundary separation, which leads to crack growth. In this paper, the high sensitivity technique for detecting creep damage (Type IV cracking), based on focused ultrasound was implemented. Signal processing algorithms were proposed that enable associating the properties of backscattered signals to the internal characteristics of the metal structure. As the result, the boundaries of the parent metal, HAZ and weld metal in an ultrasonic image can be indicated. The approach was validated on the samples exposed to different levels of creep damage and verified with metallographic analysis. It demonstrated that technique allowed detecting clouds of early stage creep damage.

An occurrence of cracks in pipework could lead to potentially very dangerous malfunction in some critical engineering systems such as power plants. There is a clear trend of replacing traditional manual testing with non-invasive in-situ methods that should detect crack formation in its early stage. Such as approach would enable replacing of unhealthy pipe components during the regular periodic outages. Ultrasonic testing is known to be a rather mature and reliable technology. However, it suffers from serious problems in detection of the cracks of subwavelength size. This paper attempts to soften aforementioned problems by investigating the influence of a duration of the unipolar excitation signal on the achieved resolution. In addition, the transducer input electricalimpedance of NDT transmitter was measured by using different excitation pulses and their levels and the results are compared with those obtained using traditional frequency sweeping method at low excitation levels. Finally, use of some advanced signal processing algorithms that might lead to the automatic detection of subwavelength voids, in scenarios with low signal-to-noise ratio, is discussed.

Bolts are one of the primary components used when constructing complex systems such as power plants, factories, railways, and similar. Due to constant stress over the years, various types of defects can appear inside bolts, making the overall structure unsafe. Detection of defective bolts can be done by employing a non-destructive material evaluation technique, such as ultrasonic testing (UT). However, the amount of data acquired during the inspection is often large, so the analysis, nowadays performed manually, lasts a long time. In this work, we propose a method based on a convolutional neural network (CNN) to classify ultrasonic scans and detect defective bolts. We propose a novel representation of the ultrasonic B-scans that we call rotational B-scans. By transforming the original database of B-scans into this novel representation, the number of images displaying a defect increases. This balances the dataset, decreases the dataset variance, and makes the training of a deep convolutional neural network significantly easier. We tested many existing architectures and based on our findings we designed a custom encoder-decoder-based classifier. Our model outperformed all the other tested models and reached an area under the receiver operating characteristic curve (AUC-ROC) of 97.4%.

Non-destructive testing is a group of methods for evaluating the integrity of components. Among them, ultrasonic inspection stands out due to its ability to visualize both shallow and deep sections of the material in the search for flaws. Testing of the critical components can be a tiring and time-consuming task. Therefore, human experts in analyzing inspection data could use a hand in discarding anomaly-free data and reviewing only suspicious data. Using such a tool, errors would be less common, inspection times would shorten and non-destructive testing would be more efficient. In this work, we evaluate multiple state-of-the-art deep-learning anomaly detection methods on the ultrasonic non-destructive testing dataset. We achieved an average performance of almost 82% of ROC AUC. We discuss in detail the advantages and disadvantages of the presented methods.

Non-destructive testing (NDT) is a set of techniques used for material inspection and detection of defects. Ultrasonic testing (UT) is one of the NDT techniques, commonly used to inspect components in the oil and gas industry, aerospace, and various types of power plants. Acquisition of the UT data is currently done automatically using robotic manipulators. This ensures the precision and uniformity of the acquired data. On the other hand, the analysis is still done manually by trained experts. Since the acquired UT data can be represented in the form of images, computer vision algorithms can be applied to analyze the content of images and localize defects. In this work, we propose a novel deep learning architecture designed specifically for defect detection from UT images. We propose a lightweight feature extractor that improves the precision and efficiency of the detector. We also modify the detection head to improve the detection of the objects with extreme aspect ratios which are common in UT images. We tested our approach on an in-house dataset with over 4000 images. The proposed architecture outperformed the previous state-of-the-art method by 1.7% (512 × 512 px input resolution) and 2.7% (384 × 384 px input resolution) while significantly decreasing the inference time.

Ultrasonic testing (UT) is one of the commonly used non-destructive testing (NDT) techniques for material evaluation and defect detection. The acquisition of UT data is largely performed automatically by using various robotic manipulators which can ensure the consistency of the recorded data. On the other hand, complete analysis of the acquired data is still performed manually by trained personnel. This makes the reliability of defect detection highly dependent on humans’ knowledge and experience. Most of the previous attempts for automated defect detection from UT data analyze individual A-scans. In such cases, valuable information present in the surrounding A-scans remains unused and limits the performance of such methods. The situation is better if a B-scan is used as an input, especially if the dataset is large enough to train a deep learning object detector. However, if each of the B-scans is analyzed individually, as it was done so far in the literature, there is still valuable information left in the surrounding B-scans that could be used to improve the precision. We showed that expanding the input layer of an existing method will not lead to an improvement and that a more complex approach is needed in order to effectively use information from neighboring B-scans. We propose two approaches based on high-dimensional feature maps merging. We showed that proposed models improve mean average precision (mAP) compared to the previous state-of-the-art model by 2% for input resolutions of $512\times 512$ pixels, and 3.4% for input resolutions of $384\times 384$ pixels.

Ultrasonic imaging is widely used for non-destructive evaluation in various industry applications. Early detection of defects in materials is the key to keeping the integrity of inspected structures. Currently, there have been some attempts to develop models for automated defect detection on ultrasonic data. To push the performance of these models even further more data is needed to train deep convolutional neural networks. A lot of data is also needed for training human experts. However, gathering a sufficient amount of data for training is a challenge due to the rare occurrence of defects in real inspection scenarios. This is why inspection results heavily depend on the inspector's previous experience. To overcome these challenges, we propose the use of Generative Adversarial Networks for generating realistic ultrasonic images. To the best of our knowledge, this work is the first one to show that a Generative Adversarial Network is able to generate images indistinguishable from real ultrasonic images. The most thorough statistical quality analysis to date of generated ultrasonic images has been conducted with the participation of human expert inspectors. The experimental results show that images generated using our Generative Adversarial Network provide the highest quality images compared to other published methods.

The analysis of the data in non-destructive ultrasonic testing of materials is a very time-intensive task. To alleviate the aforementioned strain on the human expert inspectors, a plethora of assisted analysis methods based on deep learning have been developed recently. However, most of these methods are based on the automated detection of flaws in A-scans and B-scans and therefore we propose a method based on the detection of flaws in C-scans that can reduce the complexity of manual detection of flaws in B-scans. The proposed method classifies each row of the C-scan based on whether it contains any flaws or not. Afterward, the positively classified rows are forwarded for further automated (and manual) inspection. The results show that the developed method significantly reduces the number of B-scans that have to be further analyzed.

Non-destructive testing is a set of techniques for defect detection in materials. While the set of imaging techniques are manifold, ultrasonic imaging is the one used the most. The analysis is mainly performed by human inspectors manually analyzing recorded images. The low number of defects in real ultrasonic inspections and legal issues considering data from such inspections make it difficult to obtain proper results from automatic ultrasonic image (B-scan) analysis. In this paper, we present a novel deep learning Generative Adversarial Network model for generating ultrasonic B-scans with defects in distinct locations. Furthermore, we show that generated B-scans can be used for synthetic data augmentation, and can improve the performance of deep convolutional neural object detection networks. Our novel method is demonstrated on a dataset of almost 4000 B-scans with more than 6000 annotated defects. Defect detection performance when training on real data yielded average precision of 71%. By training only on generated data the results increased to 72.1%, and by mixing generated and real data we achieve 75.7% average precision. We believe that synthetic data generation can generalize to other challenges with limited datasets and could be used for training human personnel.

Nondestructive evaluation (NDE) is a set of techniques used for material inspection and defect detection without causing damage to the inspected component. One of the commonly used nondestructive techniques is called ultrasonic inspection. The acquisition of ultrasonic data was mostly automated in recent years, but the analysis of the collected data is still performed manually. This process is thus very expensive, inconsistent, and prone to human errors. An automated system would significantly increase the efficiency of analysis, but the methods presented so far fail to generalize well on new cases and are not used in real-life inspection. Many of the similar data analysis problems were recently tackled by deep learning methods. This approach outperforms classical methods but requires lots of training data, which is difficult to obtain in the NDE domain. In this work, we train a deep learning architecture EfficientDet to automatically detect defects from ultrasonic images. We showed how some of the hyperparameters can be tweaked in order to improve the detection of defects with extreme aspect ratios that are common in ultrasonic images. The proposed object detector was trained on the largest dataset of ultrasonic images that was so far seen in the literature. In order to collect the dataset, six steel blocks containing 68 defects were scanned with a phased-array probe. More than 4000 VC-B-scans were acquired and used for training and evaluation of EfficientDet. The proposed model achieved 89.6% of mean average precision (mAP) during fivefold cross validation, which is a significant improvement compared to some similar methods that were previously used for this task. A detailed performance overview for each of the folds revealed that EfficientDet-D0 successfully detects all of the defects present in the inspected material.

Ultrasonic testing (UT) is a commonly used approach for inspection of material and defect detection without causing harm to the inspected component. To improve the reliability of defect detection, the material is often scanned from various angles leading to an immense amount of data that needs to be analyzed. Some of the defects are only seen on B-scans taken from a particular angle so discarding some of the data would increase the risk of not detecting all of the defects. Recently there has been significant progress in the development of methods for automated defect analysis from the UT data. Using such methods the inspection can be performed quicker, but it is still necessary to inspect all of the angles to detect defects. In this work, we test a novel approach for accelerating the analysis by merging the images from various angles. To reduce the information loss during the process of merging, we develop a new model with a weighting module that dynamically determines the importance of each of the scanning angles. Using the proposed module, the loss of information is minimal, so the precision of the detection model is comparable to the model tested on each of the images separately. Using the merged images input, the analysis can be accelerated by almost 15 times.

Analysis of ultrasonic testing (UT) data is a time-consuming assignment. In order to make it less demanding we propose an approach based on a variational autoencoder (VAE) to filter out the scans without anomalies/defects and in doing so, partially automate the procedure. The implemented approach uses an additional encoder network allowing to encode the reconstructed images. The differences in encodings of input and reconstructed images have shown to be good indicators of anomalous data. Anomaly detection results surpass the results of other VAE based anomaly criteria.

Non-destructive ultrasonic analysis of materials is a method for assessing the integrity of the inspected components. It is commonly used in monitoring critical parts of the power plants, in aeronautics, oil and gas, and the automotive industry. Since most ultrasonic inspections rely on expert's previous experience they must constantly practice on new, unseen data. Acquiring enough data for training human experts on non-destructive ultrasonic scan analysis can be an expensive and time-consuming task. The only possibility to get new data for practicing is to implant synthetic defects in real metal blocks. Artificial defects are made by temperature strain, electrical discharge, and physical damage. All of those methods are very complicated and expensive to perform. Also metal blocks have to be taken from the components of the power plants to have the same structure and be realistic. In this work, some attempts have been made to generate 3D ultrasonic scans using computer vision and deep learning methods.

This work presents an experimental method for obtaining the best excitation pulse type and width for optimal driving of an industrial non-destructive testing (NDT) ultrasound transducer by using its electrical impedance parameters as input. Within the optimization method, a classic low-voltage frequency sweeping signal and two different types of high-voltage impulse excitation signals (unipolar and bipolar), with different pulse widths, are used for the electromechanical characterization of the transducer. The optimization of the excitation signal is verified by measuring the voltage, current and electrical power of the excitation and received signals obtained by ultrasound pulse reflection from two cracks (2 mm and 3 mm) in a stainless-steel block. Additionally, the influence of the output electrical impedance of the power amplifier on the optimal transfer of electrical power from the amplifier to the loading (the NDT transducer on the block) and vice versa is also discussed in this work. The optimum working point for the ultrasound NDT transducer, regarding the impulse excitation type and its pulse width, is a bipolar impulse excitation with the width and amplitude calculated from maximum values of the measured voltage, electric current and power generated in the transducer receive mode by the reflected ultrasound signals.

Non-destructive ultrasonic testing (UT) of materials is used for monitoring critical parts in power plants, aeronautics, oil and gas industry, and space industry. Due to a vast amount of time needed for a human expert to perform inspection it is practical for a computer to take over that task. Some attempts have been made to produce algorithms for automatic UT scan inspection mainly using older, non-flexible analysis methods. In this paper, two deep learning based methods for flaw detection are presented, YOLO and SSD convolutional neural networks. The methods' performance was tested on a dataset that was acquired by scanning metal blocks containing different types of defects. YOLO achieved average precision (AP) of 89.7% while SSD achieved AP of 84.5%.