Battery Pouch Cell Defect Detection System Using YOLOv11 Lightweight Model `Pouch Cell Defect Detection using Lightweight YOLOv11`

The defect identification process within the UK rail industry has seen significant improvements over the past decade with the introduction of new measurement systems and defect detection systems. Although significant work has been on the defect identification little work has been done on the process after the defect has been detected. This repair process is still extremely manual. Due to the current process being manual the repair operation has very little traceability and transparency. This paper has therefore presented the need for not only a defect detection system but a defect repair system for the UK railway industry. Further to this, this paper has acknowledged that the rise of defects occurring on the UK railway lines requires a solution that can fully repair a defect with little to no user intervention in a timely manner. To address this, this paper has taken the extremely manual process of rail repair and has laid out the possibilities to automate this process. By doing this a work flow diagram has been generated to show how the system could be used to repair surface defects with a specific focus being made on squat defects. To achieve this a defect detection and measurement system has been explored, as this will make up the first stage of the automated repair system. The literature on various defect detection algorithms was reviewed and two variations of existing defect detection algorithms were created, i.e. the Covariance method and the Normal Intersection method. These algorithms have been tested against 100 simulated squat defects and have been verified using 4 experimentally generated defects. Both algorithms have been proven to not only identify the approximate size of the defect but also its location. This successful defect identification will be integrated into an automated rail repair system.

The identification and location of critical defects inside battery cells before the performance decreases or safety issues arise remain a challenge. This study compares two nondestructive testing methods for the 3D visualization of defects at different depths inside a pouch battery cell: scanning acoustic microscopy (SAM) and X‐ray computed tomography (CT). A manufactured pouch cell with eight electrode sheets is used for this investigation. SAM using a 15 MHz transducer in reflection mode can detect defects at depths of up to four electrode sheets with a lateral resolution of 150 μm in 2 min. CT can locate defects on all eight stacked electrode sheets inside the pouch cell. The CT measurements take about 12.5 h. Both methods can complement each other in detecting defects inside thin pouch cells as an end‐of‐line test after the production or for qualifying individual battery cells for second‐life applications. As an in‐line quality check, SAM has proven to be a cost‐effective and efficient method for detecting defects such as misalignment on stacked electrodes. Both methods have the potential to expand the portfolio of nondestructive quality assurance tests in the production of lithium‐ion battery cells. This contributes to increasing the safety and productivity of battery technology.

Radiographic testing is generally used in the quality management of aeroengine turbine blades. Traditional radiographic testing is critically dependent on artificially detecting professional inspectors. Thus, it sometimes tends to be error-prone and time-consuming. In this study, we gave an automatic defect detection method by combining radiographic testing with computer vision. A defect detection algorithm named DBFF-YOLOv4 was introduced for X-ray images of aeroengine turbine blades by employing two backbones to extract hierarchical defect features. In addition, a new concatenation form containing all feature maps was developed which play an important role in the present defect detection framework. Finally, a defect detection and recognition system was established for testing and output of complete turbine blade X-ray images. Meanwhile, nine cropping cycles for one defect, flipping, brightness increasing and decreasing were applied for expansion of training samples and data augmentation. The results found that this defect detection system can obtain a recall rate of 91.87%, a precision rate of 96.7%, and a false detection rate of 7% within the score threshold of 0.5. It was proven that cropping nine times and data augmentation are extremely helpful in improving detection accuracy. This study provides a new way of automatic radiographic testing for turbine blades.

The development of effective recycling technologies is essential for the recovery and reuse of the raw materials required for lithium‐ion batteries (LIBs). Future recycling processes depend on accessible information, necessitating the implementation of a digital battery passport. The European battery regulation mandates the use of a machine‐readable identifier physically attached to the batteries for accessing digital information. Since externally applied optical labels are vulnerable to mechanical damage, technologies for identification without these restrictions could be beneficial. This study demonstrates that magnetic supraparticles (SPs) can be used for contactless identification of lithium nickel manganese cobalt oxide (NMC) battery pouch cells via magnetic particle spectroscopy (MPS) and that multiple pouch cells can be discriminated based on their specific magnetic code. A comparison of three independent model scenarios revealed that the detection of SPs and the impact on cell performance are dependent on the integration location. The results validate the concept of magnetic identification in metallic environments with MPS as an alternative to optical labeling methods. This study provides a foundation for the development of a new selective labeling and identification technology for batteries, with the potential to facilitate recycling and contribute to a more sustainable future.

A high energy X-ray diffraction technique is employed in a new way to make operando through-thickness measurements inside a large format commercial Li-ion pouch cell. The technique, which has a sub-mm in-plane spatial resolution, simultaneously determines the local temperature, the local state of charge of both electrodes (as opposed to the global average state of charge determined electrochemically), and the local in-plane elastic strain in the current collectors, all without embedding any intrusive sensors that alter battery behavior. As both thermal strain and mechanical strain develop during the charge-discharge cycling of the pouch cell, a novel approach developed herein makes it possible to separate them, allowing for measurement of the local temperature inside the battery. The operando experiment reveals that the temperature inside the cell is substantially higher than the external temperature. We propose that mechanical strain is due primarily to load transfer from the electrode to the current collector during lithiation, allowing determination of the local binder. Detailed local SOC mapping illustrates non-uniform degradation of the battery pouch cell. The possibility for 3D measurements is proposed. We believe that this new approach can provide critically needed data for validation of detailed models of processes inside commercial pouch cells.

Lithium-ion battery (LIB) is a broadly adopted technology for energy storage. With increasing demands to improve the rate capability, cyclability, energy density, safety, and cost efficiency, it is crucial to establish an in-depth understanding of the detailed structural evolution and cell-degradation mechanisms during battery operation. Here, we present a laboratory-based high-resolution and high-throughput X-ray micro-computed laminography approach, which is capable of in situ visualizing of an industry-relevant lithium-ion (Li-ion) pouch cell with superior detection fidelity, resolution, and reliability. This technique enables imaging of the pouch cell at a spatial resolution of 0.5 μm in a laboratory system and permits the identification of submicron features within cathode and anode electrodes. We also demonstrate direct visualization of the lithium plating in the imaged pouch cell, which is an important phenomenon relevant to battery fast charging and low-temperature cycling. Our development presents an avenue toward a thorough understanding of the correlation among multiscale structures, chemomechanical degradation, and electrochemical behavior of industry-scale battery pouch cells.

Surface defect detection is crucial in maintaining product quality across various industries. Traditional manual inspection methods are often time-consuming and subjective, which can result in inaccuracies and higher production costs. With the use of deep learning techniques, significant advancements have been made in automating the process of surface defect detection in recent years. Moreover, deep learning includes a variety of techniques, and image recognition-based deep learning is especially relevant to our field of study, which is the main focus of this research paper.In the industrial surface defect detection field, researchers have always aimed to create a deep learning-based intelligent defect detection system that achieves near-zero defect rates while maintaining a lightweight, efficient, and cost-effective solution. However, these objectives often conflict with each other, and it is unrealistic to develop a model that can achieve all of them simultaneously. Some trade-offs must be made. If accuracy is the top priority, a large amount of defective data labeled for supervised learning is usually required. If lightweight and low cost is prioritized, a simple small model such as Auto-Encoder is usually used, along with a large number of flawless images for unsupervised learning to minimize the cost of labeling.As mentioned before, it is very difficult to design a single model that can achieve all of them simultaneously. However, present-day studies frequently center on accomplishing those tasks using a single model and rarely address the multi-model architecture. This paper presents a Surface Defect Detection and Classification System that builds on the current state-of-the-art model in the field of surface defect detection, along with the zero-shot learning (ZSL) classifier based on VAEGAN and the Variational Auto-Encoder developed by our laboratory.We have developed a Surface Defect Detection and Classification System that effectively integrates the aforementioned three models. It has been validated on a dataset of metal surface defects, yielding excellent results. This system not only achieves defect detection rates that comply with industrial standards and low false positive rates but also maintains characteristics such as lightweight, low latency, and low annotation cost. In addition to achieving the above goals, this system can also instantly identify and issue anomaly notifications when there are unseen anomalies, which is generally impossible to do with supervised learning anomaly detection models.

Digital radiography is a promising non-destructive testing tool for powder metallurgy (PM) parts, in which transmitted X-rays are recorded to generate data for an advanced defect detection system. An important part of this system is the data processing platform for pattern recognition in X-ray images. Combinations of advanced techniques for noise reduction, contrast enhancement and image segmentation are employed. Algorithms of registration for images in regions of interest are discussed, e.g. the scale invariant feature transform (SIFT). Modern pattern recognition methodologies such as smoothing, moment representation, image alignment and optical flow towards feature classification are evaluated. The proposed defect detection and classification capability for automatic analysis of digital radiographic images from PM parts potentially allows integration into multiple-view inspection systems, which should enhance quality control in the PM manufacturing and production environment. Defect detection systems able to work at the speed of current production lines are of great interest to both PM manufacturers and users.

In the process of steel production, the defects on the surface of steel will adversely affect the subsequent processing of a product. Accurate detection of such defects is the key to improve production efficiency and economic benefits. In this paper, an end-to-end steel surface defect detection and size measurement system based on the YOLOv5 model is designed. Firstly, in consideration of the defect location and direction correlation in the production process, a coordinate attention mechanism is added at the head of YOLOv5 to strengthen the spatial correlation of the steel surface and an adaptive anchor box generation method based on defect shape difference feature is proposed, which realizes the detection of three main types of defects on the Pytorch deep learning framework. Secondly, BiFPN is used to strengthen the feature fusion and a transformer encoder is added to improve the performance of detecting small defects. Thirdly, calculate the conversion ratio between the pixel and the actual size according to the standard reference specimen and obtain the actual size through the pixel statistics of the defect area to achieve pixel level size measurement. Finally, the steel surface defect detection and size measurement system are designed in this paper, which consist of various hardware, related measurement, and detection algorithms. According to the experimental results, the comprehensive defect detection accuracy of this method reaches 93.6%, of which the scratch detection accuracy reaches 95.7%. The detection speed reaches 133 fps and the defect size measurement accuracy reaches 0.5 mm. Experimental result shows that the defect detection and size measurement system designed in this paper can accurately detect and measure various industrial production defects and can be applied to the actual production process.

Impedance measurements are a powerful tool to investigate interfaces in lithium-ion batteries (LIBs). In order to deconvolute the anode and cathode contributions to the cell impedance, a reference electrode (RE) is required. However, there are only very few reports on the use of a three-electrode setup with an RE for all-solid-state batteries (ASSBs), which is due to the complexity of integrating an RE with a suitable geometry into the typical ASSB test cells that are based on a compressed electrolyte pellet. In this study, we present a straightforward approach to implement a micro-reference electrode (μ-RE) for electrode-resolved impedance and potential measurements into ASSB pouch cells. The μ-RE consists of an insulated ∼64 μm diameter gold wire that is sandwiched between two Li6PS5Cl/polymer separator sheets and activated by in situ electrochemical lithiation. Using this μ-RE, we investigate the electrode potential and the accessibility of cyclable lithium at the separator interface of indium-lithium anodes, which are prepared by stacking lithium and indium foils with a molar excess of indium. We compare two different cell assembly configurations, with the separator faced by either (i) the formerly In-side or (ii) the formerly Li-side, showing that only the latter case provides a reservoir of cyclable lithium.

Hybrid or all-electric aircraft are being developed as the next generation of aircraft to both allow new forms of aviation and decrease environmental impact. Since these types of aircraft are based on high-capacity battery technology, safe operation of these batteries becomes increasingly important. In particular, the potential for battery failure due to uncontrolled chemical reactions resulting in thermal runaway, catastrophic failure, and battery fires must be addressed in order for such battery technology to have the level of safety needed for standard aviation implementation. Efforts to ensure battery safety often involve engineering solutions that seek to contain rather than prevent such events by early detection. Such approaches increase the system weight and decrease the power per unit mass provided by the battery system. Existing methods for measuring battery parameters to determine the battery state-of-health are limited. These methods include electrical measurements of the cell current and/or voltage output as well as temperature measurements taken externally on the cell surface. Such external temperature measurements are limited in their ability to provide early warning of impending battery failure. In response, an effort to develop sensors operating internal to battery for health monitoring has been ongoing in the NASA Sensor-based Prognostics to Avoid Runaway Reactions & Catastrophic Ignition (SPARRCI) project [1]. The basic approach associated with this sensor work is the deposition of thin film sensors on the battery separator located between the anode and cathode of the battery. These thin film sensors are then monitored to determine changes in battery parameters and health. Microfabrication techniques are employed to minimize the overall impact of the sensors on battery operation through the implementation of sensors with minimal size, weight, and power consumption. The thickness of the films, which are fabricated through physical vapor deposition (sputtering), are on the order of thousands of angstroms and can have minimal surface area. Thin film sensors for system health management have been implemented for a many decades on complex components for aerospace applications [2,3]. However, the application of thin films of this type on a battery separator for internal battery monitoring applications has not previously been demonstrated to our knowledge. This paper describes the development of sensors for the internal battery monitoring through the use of thin film sensor technology. Thin metal films were successfully deposited on a battery separator polymer material with good adherence and electrical continuity. Multiple types of sensors have been deposited, as well as lead connections from the sensor to the edge of the separator material. The ability of these thin film sensors immersed in electrolyte to perform multiple types of battery parameter measurements has been demonstrated. For example, a multiparameter sensor system measured multiple properties simultaneously inside of a pouch cell over a wide temperature range. Further, real time measurement of interior temperature changes in a battery pouch cell with an integrated interior temperature sensor was demonstrated. These changes include detecting a fault in the battery (shorting) in situ with rapid response time (less than a minute) corresponding to a more limited response by a temperature sensor mounted externally. Other aspects of monitoring battery health were also explored, such as real-time measurement of simulated dendrite growth/metal deposition by sensor on separator material demonstrated. Future efforts will include improvements in the durability of the sensor structure to allow introduction of the approach into standard battery fabrication techniques. Overall, this work is a step forward in providing a method to prevent catastrophic battery failures and provide a foundation for safer, lighter, and higher energy batteries for the electric aircraft industry.[1] B. DeMattia, Daniel Perey, John Lawson, and Gary Hunter, “Advanced Battery Health Approaches for Electric Aircraft”, Energy & Mobility Technology, Systems, and Value Chain Conference & Expo, Cleveland, OH, Sept. 23, 2023.[2] John D. Wrbanek, and Gustave C. Fralick, “Thin Film Physical Sensor Instrumentation Research and Development at NASA Glenn Research Center”, 52nd International Instrumentation Symposium Cleveland, OH, May 2006, NASA TM-2006-214395[3] Lawrence G. Matus (2015) “Instrumentation for Aerospace Applications: Electronic-Based Technologies”, Journal of Aerospace Engineering 26 (2) https://doi.org/10.1061/(ASCE)AS.1943-5525.0000302

Electrochemical impedance spectroscopy (EIS) is a powerful and versatile tool to investigate interfaces in batteries. In order to disentangle the anode and cathode contributions from the full-cell impedance, a reference electrode (RE) is required. In the field of batteries based on liquid electrolytes, the concept of a RE has become a widespread tool for the EIS analysis of small-scale cells [1]. However, there are only very few reports on the use of a three-electrode setup with a reference electrode for all-solid-state batteries (ASSBs) [2,3], which is due to the complexity of integrating a RE with a suitable geometry in the typical ASSB test cells that are based on a compressed electrolyte pellet (further on referred to as bulk-type ASSB cells), since for artifact-free single-electrode impedance spectra, the RE should be placed between the electrodes and should be thin compared to the thickness of the pellet. In contrast to the widely used bulk-type ASSB cells, a recently available alternative construction is offered by the use of free-standing separator sheets based on a solid electrolyte / polymer binder composite (further on referred to as sheet-type ASSB cells),[4,5] in which case a micro-RE can be placed between two separator sheets, in analogy to the micro-RE concept used with batteries based on liquid electrolytes [1].In this study, we use sheet-type separators based on a composite consisting of Li6PS5Cl (LPSCl) solid electrolyte and a hydrogenated nitrile butadiene rubber (HNBR) binder to build ASSB pouch cells that include a gold wire micro-RE (µ-GWRE). We show that upon in-situ lithiation of the µ-GWRE a stable reference potential is obtained and that artifact-free single-electrode impedance spectra can be obtained, analogous to what we had found previously for a µ-GWRE in a lithium ion battery with liquid electrolyte.[1] Figure 1 shows both half-cell impedance spectra of an InLi | separator sheet | Li cell. The sum of both half-cell impedances (blue) is identical to the full-cell impedance (green) and now impedance loops or other common artefacts are observed for the InLi (black) and the Li (red) electrodes, indicating the viability of this setup to determine single-electrode impedances. Since the InLi electrode is commonly used as counter electrode (CE) for ASSB testing cells, we will also use this setup to investigate the potential stability of InLi alloys and their impedance evolution upon lithiation and delithiation. Acknowledgements: This work was carried out as part of the research project “Industrialisierbarkeit Festkörperelektrolytzellen”, funded by the Bavarian Ministry of Economic Affairs, Regional Development and Energy.

During cycling and high temperature storage, conventional lithium-ion electrolytes degrade through mechanisms that can lead to insoluble electrode surface films, solution-phase products, and gaseous species. Gas degradation products can cause significant cell swelling, leading to performance and safety problems. Innovations in electrolyte materials that can reduce gassing are highly desired, and gaining a greater understanding of the mechanisms of gas generation can further inform electrolyte development and optimization for reduced gassing.Fluorinated organosilicon (OS) compounds developed by Silatronix® have been studied in lithium-ion batteries, and their inclusion in the electrolyte has been shown to dramatically reduce gassing, as well as increasing cycling stability in nickel-rich battery systems.1 In this prior study, the gas species were identified and quantified, but the mechanistic origin of each gas remained unknown, Therefore, the mechanism of gas reduction by OS, as well as the effects of varying the OS molecule structure, required more investigation.In this study, isotopic labeling of multiple electrolyte components is used to provide insight into the origin of gas products, and the nature of gas reduction by several different OS materials. In a previous study, Silatronix® used 13C-labeled ethylene carbonate (EC) to identify the gases produced from EC during the battery’s first charge, as well as during high temperature storage.2 This follow-up investigation expands the 13C-labeled electrolyte components to include diethyl carbonate (DEC) and dimethyl carbonate (DMC) in addition to EC. Tertiary carbonate blends are used, each containing one 13C-labeled carbonate solvent blended with two unlabeled carbonates. Carbonate-only electrolytes are also compared to electrolytes with several different OS compounds added to the labeled carbonate formulations. NMC811/Gr pouch cells are charged and stored at high temperature, followed by gas generation measurements. GC-MS is used to analyze the gas components, labeled and unlabeled, generated in the battery pouch cells. Gases coming from EC, DEC, and DMC are each identified and quantified. The results show that all OS-containing cells have significant reductions in gas volume (-63%) compared to the carbonate-only cells, primarily through decreasing CO2. The quantification of gases shows that the largest fraction of solvent-generated CO2 comes from EC. All labeled gas species and quantities originating from each labeled carbonate are presented, and the effects of OS on the gases and their source carbonate components are shown, as well as the effect of different OS structures on gas reduction. Overall, this study illuminates in detail the origins of gassing from each of three common carbonate solvents when combined in a complex electrolyte formulation, and demonstrates that organosilicon materials are effective gas-reducing electrolyte components.

The development of Photovoltaic (PV) technology has paved the path to the exponential growth of solar cell deployment worldwide. Nevertheless, the energy efficiency of solar cells is often limited by resulting defects that can reduce their performance and lifespan. Therefore, it is crucial to identify a set of defect detection approaches for predictive maintenance and condition monitoring of PV modules. This paper presents a comprehensive review of different data analysis methods for defect detection of PV systems with a high categorisation granularity in terms of types and approaches for each technique. Such approaches, introduced in the literature, were categorised into Imaging-Based Techniques (IBTs) and Electrical Testing Techniques (ETTs). Although several review papers have investigated recent solar cell defect detection techniques, they do not provide a comprehensive investigation including IBTs and ETTs with a greater granularity of the different types of each for PV defect detection systems. Types of IBTs were categorised into Infrared Thermography (IRT), Electroluminescence (EL) imaging, and Light Beam Induced Current (LBIC). On the other hand, ETTs were categorised into Current-Voltage (I-V) characteristics analysis, Earth Capacitance Measurements (ECM), Time Domain Reflectometry (TDR), Power Losses Analysis (PLA), and Voltage and Current Measurements (VCM). Approaches based on digital/signal processing and Machine Learning (ML) models for each method are included where relevant. Moreover, the paper critically analyses the advantages and disadvantages of each of the adopted techniques, which can be referred to by future studies to identify the most suitable method considering the use-case’s requirements and setting. The adoption of each of the reviewed techniques depends on several factors, including the deployment scale, the targeted defects for detection, and the required location of defect analysis in the PV system, which are expanded further in the presented analysis. From a high-level perspective, while IBTs provide a high-resolution visual representation of the module surface, allowing for the detection and diagnosis of small structural defects that may be missed by other techniques, ETTs can detect electrical faults beyond the PV module’s surface. On the IBT level, the most notable adopted techniques in the literature are IRT- and EL-based. While IRT techniques are more practical for large-scale applications than EL imaging, the latter is considered a non-intrusive technique that is highly efficient in localising defects of solar cells. The paper also discusses challenges observed in the state-of-the-art related to data availability, real-time monitoring, accurate measurements, computational efficiency, and dataset distribution, and reviews data pre-processing and augmentation approaches that can address some of these challenges. Furthermore, potential future orientations are identified, addressing the limitations of PV defect detection systems.

According to the demand of PET bottle defect detection in actual production, the framework of the whole detection system is analyzed, and according to the production demand, each module of the PET bottle defect detection and control system is designed. Firstly, the hardware part of the detection system is designed. The hardware system is divided into several subsystems, such as mechanical transmission system, image acquisition system and bottle removal system. Through the establishment of the hardware platform, the system can realize the functions of automatic image collection, automatic analysis and judgment, and automatic elimination of unqualified bottles. Secondly, SIFT algorithm is used to design the software system to achieve accurate image matching and automatic detection of PET bottle defects. Finally, the system is applied to the actual PET bottle production line, the research and experimental results show that: the system for PET bottle defect detection accuracy of 99%.