Tomography analysis tool: an application for image analysis based on unsupervised machine learning

More than a century after the discovery of superconductors (SCs), numerous studies have been accomplished to take the advantage of SCs in physics, power engineering, quantum computing, electronics, communications, aviation, health care, and defence-related applications. However, there are still challenges that hinder the full-scale commercialization of SCs, such as the high cost of superconducting wires/tapes, technical issues related to AC losses, the structure of superconducting devices, the complexity and high cost of the cooling systems, the critical temperature, and manufacturing related issues. In the current century, massive advancements have been achieved on artificial intelligence (AI) techniques by offering disruptive solutions to handle engineering problems. Consequently, AI techniques can be implemented to tackle those challenges facing superconductivity and act as a shortcut towards full commercialization of SCs and their applications. AI approaches are capable of providing fast, efficient, and accurate solutions for the technical, manufacturing, and economic problems with a high level of complexity and nonlinearity in the field of superconductivity. In this paper, concept of AI and the widely used algorithms are first given. Then a critical topical review is presented for those conducted studies that used AI methods for improvement, design, condition monitoring, fault detection and location of superconducting apparatuses in large scale power applications, as well as the prediction of critical temperature and the structure of new SCs, and any other related applications. The topical review are presented in three main categories, AI for large-scale superconducting applications, AI for superconducting materials, and AI for physics of superconductors. In addition, the challenges to apply AI techniques to the superconductivity and its applications are given. Eventually, future trends on how to integrate AI techniques with superconductivity towards the commercialization are discussed.

Section: Insulation Goodness For Coil Winding Magnets and Cables Duri...mentioning

confidence: 99%

Artificial intelligence methods for applied superconductivity: material, design, manufacturing, testing, operation, and condition monitoring

Yazdani-Asrami

Sadeghi

Song

et al. 2022

“…The equivalent stress is calculated by dividing the applied force by the groove dimension. At 19 T after unloading the force, the measured permanent I c degradation was 15% of the initial I c , and a reduction of B c2 of ∼0.5 T was also recorded [37]. The latter information indicates that plastic deformation of the copper matrix is present and plays a relevant role in the degradation of the electrical performance of this specific RRP wire.…”

Section: Samples Preparationmentioning

confidence: 70%

“…The cracks do not reduce the effective Nb 3 Sn cross section, therefore the plastic deformation of the copper and the consequent B c2 reduction must be the major responsible for the critical current irreversible decrease in this type of wire when submitted to transverse loads. In a future publication, which is under preparation, it will be proven that the main reason for the I c irreversible degradation on RRP wires is the copper plasticization [37]. These residual stresses are already relevant above 140 MPa, while cracks appear at far higher pressures in the range 200-250 MPa.…”

Section: Discussionmentioning

confidence: 87%

“…Cracks and voids were segmented from these images to produce the output examples. The cracks were manually detected one by one using GNU Image Manipulation Program [39], while the voids were segmented exploiting the TAT [31]. TAT is a machine learning application that uses pixel colour to segment tomographic images into clusters.…”

Section: U-net Training and Validationmentioning

confidence: 99%

See 1 more Smart Citation

Formation and propagation of cracks in RRP Nb₃Sn wires studied by deep learning applied to x-ray tomography

Bagni

Mauro

Majkut

et al. 2022

Self Cite

This paper reports a novel non-destructive and non-invasive method to investigate crack formation and propagation in high-performance Nb3Sn wires by combining X-Ray tomography and deep learning networks. The next generation of high field magnet applications relies on the development of new Nb3Sn wires capable to withstand the large stresses generated by Lorentz forces during magnets operation. These stresses can cause a permanent reduction of the transport properties generated by residual deformation of the Nb3Sn crystal lattice as well as the formation of cracks in the brittle Nb3Sn filaments. Studies for the development of the High Luminosity LHC (HL-LHC) upgrade showed that nominal transverse compressive stresses above 150 MPa may be sufficient to generate cracks in the wires. In the case of fusion magnets, wires experience periodic bending due to the electro-magnetic cycles of the reactor which over time may induce wire deformation and filament cracks. Therefore, it has become essential to develop a quantitative method for the characterization of crack formation and propagation under compressive loads. The X-Ray tomographic data of a series of Restacked-Rod-Process (RRP) Nb3Sn wires was acquired at the microtomography beamline ID19 of the European Synchrotron Radiation Facility (ESRF), after intentionally inducing a broad spectrum of cracks in the Nb3Sn sub-elements. The samples were submitted to transvers compressive stresses, with and without epoxy impregnation, at different pressures, up to 238 MPa. The resulting tomographic images were analysed by means of deep learning semantic segmentation networks, using U-net, a Convolutional Neural Network (CNN), to identify and segment cracks inside the wires. The trained CNN was able to analyse large volumes of tomographic data, thus enabling a systematic approach for investigating the mechanical damages in Nb3Sn wires. We will show the complete three-dimensional (3D) reconstruction of various cracks and discuss their impact on the electro-mechanical performance of the analysed wires.

“…The wire volume examined by tomography corresponds to a scan length of about 1.5 mm. More details about the methods of detection and image recognition are reported in the [29,33,34]. Only very few cracks are present in the examined sample after unload from 240 MPa.…”

Section: Resultsmentioning

confidence: 99%

Degradation of I _c due to residual stress in high-performance Nb₃Sn wires submitted to compressive transverse force

Senatore

Bagni

Troitino

et al. 2023

Self Cite

Future particle colliders in search for new physics at the energy frontier require the development of accelerator magnets able of producing fields well beyond those attainable with Nb-Ti. As the next generation of high field accelerator magnets is presently planned to be based on Nb3Sn, it becomes crucial to establish precisely the mechanical limits at which this brittle and strain sensitive superconductor can operate safely. This paper reports on the stress dependence and the permanent reduction of the critical current under transverse compressive loads up to 240 MPa in state-of-the-art Restacked-Rod-Process (RRP®) and Powder-In Tube (PIT) Nb3Sn wires. Single-wire experiments were performed at 4.2 K in magnetic fields ranging between 16 T and 19 T on resin-impregnated samples to imitate the operating conditions of a wire in the Rutherford cable of an accelerator magnet. Depending on the wire technology, we measured irreversible stress limit values – defined as the transverse stress value leading to a permanent reduction in the critical current of 5%, assessed by convention at 19 T – ranging between 110 MPa and 175 MPa. This permanent reduction of the critical current after mechanical unload can occur for two reasons, which can be concomitant: the plastic deformation of the Cu matrix that produces residual stresses on the Nb3Sn lattice and the formation of cracks. We developed a method to identify the dominant degradation mechanismin our experiments that allowed us predicting the fraction of critical current lost due to the residual stresses. Interestingly, we found that in the RRP® wires the measured reduction of Ic after unload from stresses as high as 240 MPa can be fully ascribed to residual stresses. An independent confirmation of this conclusion coming from a study combining X-Ray tomography and deep learning Convolutional Neural Networksis also reported.