Refine
Year of publication
Document Type
- Article (223)
- Conference Proceeding (200)
- Part of a Book (31)
- Other (31)
- Book (13)
- Preprint (12)
- Report (4)
- Doctoral Thesis (3)
- Course Material (2)
- Contribution to a Periodical (1)
Language
- English (521) (remove)
Has Fulltext
- no (521) (remove)
Keywords
- Dissipative Particle Dynamics (4)
- OCSR (3)
- Polymer-Elektrolytmembran-Brennstoffzelle (3)
- CDK (2)
- DECIMER (2)
- Deep Learning (2)
- Deep learning (2)
- Electrodeposition (2)
- Erweiterte Realität <Informatik> (2)
- Field measurement (2)
- OCSR, Optical Chemical Structure Recognition (2)
- Solar modules (2)
- Transformer (2)
- Twitter <Softwareplattform> (2)
- hydraulic compression (2)
- modular stack design (2)
- open quantum systems (2)
- water electrolysis (2)
- carbon nanofibers, platinum electrodeposition, ele ctrochemical surface area (1)
- 360 degree Feedback (1)
- AEM-Electrolysis (1)
- AI (1)
- Additive manufacturing Directed energy deposition-arc 316L stainless steel Corrosion behavior Electrochemical corrosion (1)
- Aerosol (1)
- Air handling unit (1)
- AlphaFold, ColabFold, PyMOL (1)
- Alternative Geschäftsmodelle (1)
- Assessment Center (1)
- Assisted living technologies (1)
- Assistive robotics (1)
- Augmented (1)
- Augmented Electromagnetic Accelerators (1)
- Augmented Multiphase (1)
- Augmented Multiphase Rail Launcher (1)
- Augmented Reality (1)
- Augmented Three-Phase AC-Railgun (1)
- Biomedical monitoring, Hospitals, Electrocardiography, Wireless communication, Patient monitoring, Wireless sensor networks (1)
- Bone Morphogenetic Protein, BMP, BMP2 (1)
- Brand theory (1)
- CFD Simulation (1)
- Carboxylate (1)
- Case-Study (1)
- Catalysis (1)
- Cell-free implant (1)
- Chemical image depiction (1)
- Chemical space (1)
- Chemical structure depictions (1)
- Cheminformatics (1)
- Chemistry Development Kit (1)
- Chemistry Development Kit, CDK, Molecule fragmentation, In silico fragmentation, Scaffolds, Functional groups, Glycosidic moieties, Rich client, Graphical user interface, GUI (1)
- Chief Executive Officer (1)
- Climate change (1)
- Clustering (1)
- Communication management (1)
- Continuous Queries (1)
- Crowdfunding (1)
- Current Pulses (1)
- DPD, Dissipative Particle Dynamics (1)
- Data Journalism (1)
- Datalog (1)
- Datenjournalismus (1)
- Datensatz (1)
- Deductive Databases (1)
- Depiction generator image augmentation (1)
- Dissipative particle dynamics, DPD, Surfactant, Bilayer, Lamellar, Simulation, Mesoscopic (1)
- Electrolysis (1)
- Electromagnetic Launcher (1)
- Elektrodenvorbereitung (1)
- Energy Storage Mode (1)
- Evidence-based Management (1)
- Flat-Channel (1)
- Flory-Huggins parameter (1)
- Fragmentation (1)
- Future capacity needs (1)
- Gas Diffusion Electrode (1)
- Greek dept crisis (1)
- Hand-drawn chemical structures (1)
- Hand-drawn images (1)
- High Reynold Numer (1)
- Homogene Kühlung (1)
- Human-Robot Interaction (1)
- Human-centered computing (1)
- Hydraulic cell compression (1)
- Hydraulic compression, Carbon Nano Fibers, PEM Fuel Cells, Catalyst utilization (1)
- Hydrogen evolution reaction (1)
- Implantat (1)
- Incremental Evaluation (1)
- Indigo (1)
- Journalismus (1)
- Juristenausbildung (1)
- Kalman filter (1)
- Kernspintomografie (1)
- Kohlenstoff (1)
- Laser Synthesis Electrocatalytic Water Splitting (1)
- Launcher (1)
- Leadership Competencies (1)
- Linear Electromagnetic Accelerator (1)
- Lüftungsanlage (1)
- MITRE (1)
- Magnetic Pressure (1)
- Media Brands (1)
- Media brand characteristics (1)
- Media positioning (1)
- Membrane (1)
- Mixed Reality (1)
- Modular Augmented Launcher (1)
- Modular Design (1)
- Molecule images (1)
- Multiphase Rail Launcher (1)
- Mund-Nasen-Schutz (1)
- Muzzle Velocity (1)
- N,N,O Ligands (1)
- N,N′ Ligands (1)
- Nanofaser (1)
- Natural products (1)
- Naturstoff (1)
- New Work, Information and Communication Industry, Innovation, Organizational Goals, Survey (1)
- Ni-Mo alloy Catalyst (1)
- NiCrBSi coatings; flame spraying; induction remelting; wear resistance (1)
- Normalisierung (1)
- ORR OER (1)
- OSINT (1)
- Optical Chemical Structure Recognition (1)
- PEM Electrolysis, Hydrogen, Hydraulic Compression, High Pressure (1)
- PEM electrolysis (1)
- PEM fuel cell (1)
- PEM fuel cell electrocatalysts, Carbon nanofibers, Oxygen plasma activation, Pulsed electroplating. (1)
- PEM fuel cells (1)
- PEM water electrolysis (1)
- PEM-Brennstoffzelle (1)
- PEMWE (1)
- People with disabilities (1)
- Performance prediction (1)
- Physics-Informed Deep Learning (1)
- Physics-informed deep learning; unsupervised learning; Reynolds-averaged Navier-Stokesequations; high Reynolds number flow; turbulence modeling (1)
- Politische Berichterstattung (1)
- Porous Transport Layers (1)
- Powder feed rate HVOF Cermet Wear Corrosion (1)
- PtCoMn (1)
- RDKit (1)
- RIS (1)
- Rail Launcher (1)
- Railgun (1)
- Robot assistive drinking (1)
- Robot assistive eating (1)
- SARS-CoV-2 (1)
- Sauerstoffplasmaaktivierung (1)
- Scaffold (1)
- Scaffold network (1)
- Scaffold tree (1)
- Segmentation; Correlation; Diseases; Convolutional Neural Networks (1)
- Semi-Infinite Plate (1)
- Sinusoidal (1)
- Social Media (1)
- Spondylodese (1)
- Stack <Brennstoffzelle> (1)
- Stellite 6; HVOF-spraying; Laser remelting; Cavitation erosion; Coatings (1)
- Temperature coefficients (1)
- Ternary alloy catalyst preparation (1)
- Tetraplegie (1)
- Thermal Performance (1)
- Thermal Stress (1)
- Titanium; Al2O3–TiO2 coatings; Nanoindentation (1)
- Twitter (1)
- Update Propagation (1)
- Upscaling laboratory models (1)
- Urban heat island (1)
- Zinc (1)
- Zustandsmaschine (1)
- academic and job-related self-control demands (1)
- additive manufacturing; polylactic acid (PLA); fused filament fabrication (FFF); fused deposition modeling (FDM); printing temperature; filament color; dimensional accuracy; tensile strength; friction performance; wear (1)
- adhesion (1)
- advanced persistent threats (1)
- aerosol (1)
- air hygiene (1)
- airborne infection (1)
- antecedents (1)
- artificial intelligence (1)
- assistive robotics (1)
- augmented reality (1)
- balance (1)
- barrel cortex, cell types, multielectrode recordings, perception, psychophysics (1)
- bio-inspired functional surface (1)
- biomimetic (1)
- bipolar plate (1)
- carbon nano fibres (1)
- carbon nanofibers, platinum electrodeposition, electrocatalysts (1)
- cartilage defect (1)
- cartilage regeneration (1)
- cluster (1)
- co-based alloys; hot corrosion; solid particle erosion; microstructure; brazing (1)
- coatings (1)
- cobot (1)
- compassionate love (1)
- conical intersection (1)
- consent banner (1)
- cookie banner (1)
- cookies (1)
- corrosion resistance (1)
- corrosion; self-fluxing alloys; NiCrBSi; WC-12Co; cavitation; hard metals (1)
- critical review (1)
- cyber kill chain (1)
- demagnetization cooling (1)
- diversity (1)
- dynamic capabilities (1)
- education; skills; competences; pandemic; online or face-to-face (1)
- efficiency of exciton transfer (1)
- electrode preparation (1)
- empowerment (1)
- entrepreneurial diversity (1)
- entrepreneurship (1)
- ethics (1)
- excitation energy transfer (1)
- expert interviews (1)
- face mask (1)
- fused deposition modeling (FDM); fused filament fabrication (FFF); polylactic acid (PLA); layer height; layer thickness; filament color; PLA color; dimensional accuracy; tensile strength (1)
- human robot interaction (1)
- human-centered design (1)
- human-robot collaboration (1)
- hybrid sensor system (1)
- hydraulic cell compression (1)
- infrared heating panel (1)
- ingots (1)
- intermolecular interaction (1)
- international comparative study (1)
- intersectionality (1)
- irritation (1)
- machine learning (1)
- managerial vs. non-managerial actors (1)
- measurement study (1)
- media accountability (1)
- microfoundations (1)
- molecular force field (1)
- multi-level model of competence (1)
- narcissism (1)
- neutrality (1)
- nonadiabatic dynamics (1)
- noncommuting fluctuations (1)
- nonequilibrium quantum transport (1)
- normalisation (1)
- novel (1)
- optical chemical structure recognition (1)
- oxygen plasma activation (1)
- participatory design (1)
- phishing (1)
- photovoltaic power systems (1)
- political journalism (1)
- polymer electrolyte membrane (1)
- privacy (1)
- projection (1)
- protein structure prediction (1)
- quality standards (1)
- quantum dissipation (1)
- reconnaissance (1)
- relevance (1)
- risk management (1)
- role identity (1)
- self-fluxing; ZrO2; NiCrBSi; vacuum post-treatment; thermal spraying (1)
- sensor fusion (1)
- servant leadership (1)
- shared user control (1)
- silicon (1)
- solar cells (1)
- spatial policy (1)
- state machine (1)
- study and working time per week (1)
- sustainable development (1)
- television news coverage (1)
- trait self-control (1)
- tree frog (1)
- user acceptance (1)
- ventilation (1)
- vibronic coupling (1)
- virtual reality (1)
- visual cues (1)
- visualization techniques (1)
- watchblogs (1)
- web measurement (1)
Institute
- Westfälisches Institut für Gesundheit (115)
- Westfälisches Energieinstitut (60)
- Institut für Internetsicherheit (55)
- Elektrotechnik und angewandte Naturwissenschaften (49)
- Wirtschaft und Informationstechnik Bocholt (46)
- Institut für biologische und chemische Informatik (44)
- Informatik und Kommunikation (29)
- Maschinenbau Bocholt (21)
- Maschinenbau und Facilities Management (13)
- Wirtschaftsingenieurwesen (13)
This paper presents a novel approach to build consistent 3D maps for multi robot cooperation in USAR environments. The sensor streams from unmanned aerial vehicles (UAVs) and ground robots (UGV) are fused in one consistent map. The UAV camera data are used to generate 3D point clouds that are fused with the 3D point clouds generated by a rolling 2D laser scanner at the UGV. The registration method is based on the matching of corresponding planar segments that are extracted from the point clouds. Based on the registration, an approach for a globally optimized localization is presented. Apart from the structural information of the point clouds, it is important to mention that no further information is required for the localization. Two examples show the performance of the overall registration.
Global registration of heterogeneous ground and aerial mapping data is a challenging task. This is especially difficult in disaster response scenarios when we have no prior information on the environment and cannot assume the regular order of man-made environments or meaningful semantic cues. In this work we extensively evaluate different approaches to globally register UGV generated 3D point-cloud data from LiDAR sensors with UAV generated point-cloud maps from vision sensors. The approaches are realizations of different selections for: a) local features: key-points or segments; b) descriptors: FPFH, SHOT, or ESF; and c) transformation estimations: RANSAC or FGR. Additionally, we compare the results against standard approaches like applying ICP after a good prior transformation has been given. The evaluation criteria include the distance which a UGV needs to travel to successfully localize, the registration error, and the computational cost. In this context, we report our findings on effectively performing the task on two new Search and Rescue datasets. Our results have the potential to help the community take informed decisions when registering point-cloud maps from ground robots to those from aerial robots.
A compact and efficient PEM electrolyser stack design based on hydraulic single cell compression
(2019)
For this study gas diffusion electrodes (GDE) with low platinum loading are prepared for the application as anode in polymer electrolyte membrane fuel cell (PEMFC) systems based on hydraulic compression. As catalyst support material, carbon nanofibers (CNF) are investigated because of their high specific surface area and high graphitization degree. The electrode preparation is optimized by an economic and environmental friendly pre-treatment process in oxygen plasma. For GDE manufacture an ink containing oxygen plasma activated CNFs as well as hydrophilic polymer is used. After spray coating of this CNF ink on a graphitic substrate, platinum is deposited using the pulse plating technique. Preliminary results showed a considerable improvement of CNF dispersibility as well as an increased amount and an optimized morphology of the deposited platinum. Morphology and microstructure are observed by scanning electron microscopy as well as transmission electron microscopy. Platinum loading is determined by thermogravimetric analysis to be in the range of 0.01 mg cm-2 to 0.017 mg cm-2. Furthermore, MEAs are prepared from these GDEs and testing is performed in a novel modular fuel cell test stack based on hydraulic compression. Technical information about stack design and functions is given in this work.
A Crypto-Token Based Charging Incentivization
Scheme for Sustainable Light Electric Vehicle
Sharing
(2021)
The ecological impact of shared light electric vehicles (LEV) such as kick scooters is still widely discussed. Especially the fact that the vehicles and batteries are collected using diesel vans in order to charge empty batteries with electricity of unclear origin is perceived as unsustainable. A better option could be to let the users charge the vehicles themselves whenever it is necessary. For this, a decentralized,flexible and easy to install network of off-grid solar charging stations could bring renewable electricity where it is needed without sacrificing the convenience of a free float sharing system. Since the charging stations are powered by solar energy the most efficient way to utilize them would be to charge the vehicles when the sun is shining. In order to make users charge the vehicle it is necessary to provide some form of benefit for
them doing so. This could be either a discount or free rides. A
particularly robust and well-established mechanism is controlling incentives via means of blockchain-based cryptotokens. This paper demonstrates a crypto-token based scheme for incentivizing users to charge sharing vehicles during times of considerable solar irradiation in order to contribute to more sustainable mobility services.
Stereotactic frame systems are widely used in neurosurgery. The accuracy of frame devices is considered as a gold standard to which the accuracy of new frameless stereotactic navigation systems is compared. The purpose of this study is to develop a general approach for the prediction of the application accuracy of stereotactic systems. The approach will be applied to the frame‐based biopsy performed with three frame devices: Leksell G, Cosman–Roberts–Wells (CRW), and Brown–Roberts–Wells (BRW). A work‐flow analysis will be carried out demonstrating that the accuracy relevant for a clinical application comprises several error sources including imaging, target and entry point selection, image to frame coordinates registration, and the setting of mechanical parameters of the frame. These error sources will be postulated to obey a Gaussian distribution probability density. The linear, i.e., Gaussian, error propagation, will be used to link all error contributions thus to calculate the cumulative accuracy of the frame used in the application. Although the Gaussian approach is an approximation, it allows for an analytical treatment of the accuracy. Both the accuracy at the target point and the accuracy of the probe needle guidance along the planned trajectory have been investigated. Of great significance is the relationship found between accuracy, pixel dimension, and image slice thickness, the latter being the dominant factor for slices of more than 1.5 mm thickness, yielding inaccuracies larger than 1.5 mm. For target points the predictions for the application accuracy have been compared to the results of measurements, showing good agreement with the experimental data.
Cookie notices (or cookie banners) are a popular mechanism for websites to provide (European) Internet users a tool to choose which cookies the site may set. Banner implementations range from merely providing information that a site uses cookies over offering the choice to accepting or denying all cookies to allowing fine-grained control of cookie usage. Users frequently get annoyed by the banner’s pervasiveness as they interrupt “natural” browsing on the Web. As a remedy, different browser extensions have been developed to automate the interaction with cookie banners.
In this work, we perform a large-scale measurement study comparing the effectiveness of extensions for “cookie banner interaction.” We configured the extensions to express different privacy choices (e.g., accepting all cookies, accepting functional cookies, or rejecting all cookies) to understand their capabilities to execute a user’s preferences. The results show statistically significant differences in which cookies are set, how many of them are set, and which types are set—even for extensions that aim to implement the same cookie choice. Extensions for “cookie banner interaction” can effectively reduce the number of set cookies compared to no interaction with the banners. However, all extensions increase the tracking requests significantly except when rejecting all cookies.
Membrane electrode assemblies (MEA) developed at the Westphalian Energy Institute for polymer electrolyte membrane fuel cells (PEMFC) are high tech systems containing various materials structured in nanoscale, at which electrochemical reactions occur on catalyst nano particle surfaces. For low reactance homogeneous compression of the MEA’s layers is necessary. A novel stack architecture for electrochemical cells, especially PEMFC as well as PEM electrolysers, has been developed according to achieve ideal cell operation conditions. Single cells of such a stack are inserted into flexible slots that are surrounded by hydraulic media. While operation the hydraulic media is pressurised which leads to an even compression and cooling of the stack’s cells. With this stack design it has been possible to construct a test facility for simultaneous characterisation of several MEA samples. As compression and temperature conditions of every single sample are equal, with the novel test system the effect of e.g. different electrode configurations can be investigated. Furthermore, the modular stack design leads to the development of hybrid energy applications combining fuel cells, electrolysers, batteries as well as metal hydride tanks in one system.
In this work, a novel polymer electrolyte membrane water electrolyzer (PEMWE) test cell based on hydraulic single-cell compression is described. In this test cell, the current density distribution is almost homogeneous over the active cell area due to hydraulic cell clamping. As the hydraulic medium entirely surrounds the active cell components, it is also used to control cell temperature resulting in even temperature distribution. The PEMWE single-cell test system based on hydraulic compression offers a 25 cm2 active surface area (5.0 × 5.0 cm) and can be operated up to 80°C and 6.0 A/cm2. Construction details and material selection for the designed test cell are given in this document. Furthermore, findings related to pressure distribution analyzed by utilizing a pressure-sensitive foil, the cell performance indicated by polarization curves, and the reproducibility of results are described. Experimental data indicate the applicability of the presented testing device for relevant PEMWE component testing and material analysis.
In this experimental work we present a novel electrolyzer system for the production of hydrogen and oxygen at high pressure levels without an additional mechanical compressor. Due to its control strategies, the operation conditions for this electrolyzer can be kept optimal for each load situation of the system. Furthermore, the novel system design allows for dynamic long-term operation as well as for easy maintainability. Therefore, the device meets the requirements for prospective power-to-gas applications, especially, in order to store excess energy from renewable sources. A laboratory scale device has been developed and high-pressure operation was validated. We also studied the long-term stability of the system by applying dynamic load cycles with load changes every 30 sec. After 80 h of operation the used membrane electrode assembly (MEA) was investigated by means of SEM, EDX and XRD analysis.
A Robust Interface for Head Motion based Control of a Robot Arm using MARG and Visual Sensors
(2018)
Head-controlled human machine interfaces have gained popularity over the past years, especially in the restoration of the autonomy of severely disabled people, like tetraplegics. These interfaces need to be reliable and robust regarding the environmental conditions to guarantee safety of the user and enable a direct interaction between a human and a machine. This paper presents a hybrid MARG and visual sensor system for head orientation estimation which is in this case used to teleoperate a robotic arm. The system contains a Magnetic Angular Rate Gravity (MARG)-sensor and a Tobii eye tracker 4C. A MARG sensor consists of tri-axis accelerometer, gyroscope as well as a magnetometer which enable a complete measurement of orientation relative to the direction of gravity and magnetic field of the earth. The tri-axis magnetometer is sensitive to external magnetic fields which result in incorrect orientation estimation from the sensor fusion process. In this work the Tobii eye tracker 4C is used to increase head orientation estimation because it also features head tracking even though it is commonly used for eye tracking. This type of visual sensor does not suffer magnetic drift. However, it computes orientation data only, if a user is detectable. Within this work a state machine is presented which enables data fusion of the MARG and visual sensor to improve orientation estimation. The fusion of the orientation data of MARG and visual sensors enables a robust interface, which is immune against external magnetic fields. Therefore, it increases the safety of the human machine interaction.
In this work a mathematical approach to calculate solar panel temperature based on measured irradiance, temperature and wind speed is applied. With the calculated module temperature, the electrical solar module characteristics is determined. A program developed in MatLab App Designer allows to import measurement data from a weather station and calculates the module temperature based on the mathematical NOCT and stationary approach with a time step between the measurements of 5 minutes. Three commercially available solar panels with different cell and interconnection technologies are used for the verification of the established models. The results show a strong correlation between the measured and by the stationary model predicted module temperature with a coefficient of determination R2 close to 1 and a root mean square deviation (RMSE) of ≤ 2.5 K for a time period of three months. Based on the predicted temperature, measured irradiance in module plane and specific module information the program models the electrical data as time series in 5-minute steps. Predicted to measured power for a time period of three months shows a linear correlation with an R2 of 0.99 and a mean absolute error (MAE) of 3.5, 2.7 and 4.8 for module ID 1, 2 and 3. The calculated energy (exemplarily for module ID 2) based on the measured, calculated by the NOCT and stationary model for this time period is 118.4 kWh, resp. 116.7 kWh and 117.8 kWh. This is equivalent to an uncertainty of 1.4% for the NOCT and 0.5% for the stationary model.
Robot arms are one of many assistive technologies used by people with motor impairments. Assistive robot arms can allow people to perform activities of daily living (ADL) involving grasping and manipulating objects in their environment without the assistance of caregivers. Suitable input devices (e.g., joysticks) mostly have two Degrees of Freedom (DoF), while most assistive robot arms have six or more. This results in time-consuming and cognitively demanding mode switches to change the mapping of DoFs to control the robot. One option to decrease the difficulty of controlling a high-DoF assistive robot arm using a low-DoF input device is to assign different combinations of movement-DoFs to the device’s input DoFs depending on the current situation (adaptive control). To explore this method of control, we designed two adaptive control methods for a realistic virtual 3D environment. We evaluated our methods against a commonly used non-adaptive control method that requires the user to switch controls manually. This was conducted in a simulated remote study that used Virtual Reality and involved 39 non-disabled participants. Our results show that the number of mode switches necessary to complete a simple pick-and-place task decreases significantl when using an adaptive control type. In contrast, the task completion time and workload stay the same. A thematic analysis of qualitative feedback of our participants suggests that a longer period of training could further improve the performance of adaptive control methods.
Under ambient conditions, almost all metals are coated by an oxide. These coatings, the result of a chemical reaction, are not passive. Many of them bind, activate and modify adsorbed molecules, processes that are exploited, for example, in heterogeneous catalysis and photochemistry. Here we report an effect of general importance that governs the bonding, structure formation and dissociation of molecules on oxidic substrates. For a specific example, methanol adsorbed on the rutile TiO2(110) single crystal surface, we demonstrate by using a combination of experimental and theoretical techniques that strongly bonding adsorbates can lift surface relaxations beyond their adsorption site, which leads to a sig- nificant substrate-mediated interaction between adsorbates. The result is a complex super- structure consisting of pairs of methanol molecules and unoccupied adsorption sites. Infrared spectroscopy reveals that the paired methanol molecules remain intact and do not depro- tonate on the defect-free terraces of the rutile TiO2(110) surface.
It is well-known that protein-modified implant surfaces such as TiO2 show a higher bioconductivity. Fibronectin is a glycoprotein from the extracellular matrix (ECM) with a major role in cell adhesion. It can be applied on titanium oxide surfaces to accelerate implant integration. Not only the surface concentration but also the presentation of the protein plays an important role for the cellular response. We were able to show that TiOX surfaces modified with biotinylated fibronectin adsorbed on a streptavidin-silane self-assembly multilayer system are more effective regarding osteoblast adhesion than surfaces modified with nonspecifically bound fibronectin. The adsorption and conformation behavior of biotinylated and nonbiotinylated (native) fibronectin was studied by surface plasmon resonance (SPR) spectroscopy and atomic force microscopy (AFM). Imaging of the protein modification revealed that fibronectin adopts different conformations on nonmodified compared to streptavidin-modified TiOX surfaces. This conformational change of biotinylated fibronectin on the streptavidin monolayer delivers a fibronectin structure similar to the conformation inside the ECM and therefore explains the higher cell affinity for these surfaces.
Advanced Determination of Temperature Coefficients of Photovoltaic Modules by Field Measurements
(2023)
In this work data from outdoor measurements, acquired over the course of up to three years on commercially available solar panels, is used to determine the temperature coefficients and compare these to the information as stated by the producer in the data sheets. A program developed in MatLab App Designer allows to import the electrical and ambient measurement data. Filter algorithms for solar irradiance narrow the irradiance level down to ~1000 W/m2 before linear regression methods are applied to obtain the temperature coefficients. A repeatability investigation proves the accuracy of the determined temperature coefficients which are in good agreement to the supplier specification if the specified values for power are not larger than -0.3%/K. Further optimization is achieved by applying wind filter techniques and days with clear sky condition. With the big (measurement) data on hand it was possible to determine the change of the temperature coefficients for varying irradiance. As stated in literature we see an increase of the temperature coefficient of voltage and a decline for the temperature coefficient of power with increasing irradiance.
Advancements in hand-drawn chemical structure recognition through an enhanced DECIMER architecture
(2024)
Accurate recognition of hand-drawn chemical structures is crucial for digitising hand-written chemical information in traditional laboratory notebooks or facilitating stylus-based structure entry on tablets or smartphones. However, the inherent variability in hand-drawn structures poses challenges for existing Optical Chemical Structure Recognition (OCSR) software. To address this, we present an enhanced Deep lEarning for Chemical ImagE Recognition (DECIMER) architecture that leverages a combination of Convolutional Neural Networks (CNNs) and Transformers to improve the recognition of hand-drawn chemical structures. The model incorporates an EfficientNetV2 CNN encoder that extracts features from hand-drawn images, followed by a Transformer decoder that converts the extracted features into Simplified Molecular Input Line Entry System (SMILES) strings. Our models were trained using synthetic hand-drawn images generated by RanDepict, a tool for depicting chemical structures with different style elements. A benchmark was performed using a real-world dataset of hand-drawn chemical structures to evaluate the model's performance. The results indicate that our improved DECIMER architecture exhibits a significantly enhanced recognition accuracy compared to other approaches.
Advancements in Hand-Drawn Chemical Structure Recognition through an Enhanced DECIMER Architecture
(2024)
Accurate recognition of hand-drawn chemical structures is crucial for digitising hand-written chemical information found in traditional laboratory notebooks or for facilitating stylus-based structure entry on tablets or smartphones. However, the inherent variability in hand-drawn structures poses challenges for existing Optical Chemical Structure Recognition (OCSR) software. To address this, we present an enhanced Deep lEarning for Chemical ImagE Recognition (DECIMER) architecture that leverages a combination of Convolutional Neural Networks (CNNs) and Transformers to improve the recognition of hand-drawn chemical structures. The model incorporates an EfficientNetV2 CNN encoder that extracts features from hand-drawn images, followed by a Transformer decoder that converts the extracted features into Simplified Molecular Input Line Entry System (SMILES) strings. Our models were trained using synthetic hand-drawn images generated by RanDepict, a tool for depicting chemical structures with different style elements. To evaluate the model's performance, a benchmark was performed using a real-world dataset of hand-drawn chemical structures. The results indicate that our improved DECIMER architecture exhibits a significantly enhanced recognition accuracy compared to other approaches.
Inspired by the super-human performance of deep learning models in playing the game of Go after being presented with virtually unlimited training data, we looked into areas in chemistry where similar situations could be achieved. Encountering large amounts of training data in chemistry is still rare, so we turned to two areas where realistic training data can be fabricated in large quantities, namely a) the recognition of machine-readable structures from images of chemical diagrams and b) the conversion of IUPAC(-like) names into structures and vice versa. In this talk, we outline the challenges, technical implementation and results of this study.
Optical Chemical Structure Recognition (OCSR): Vast amounts of chemical information remain hidden in the primary literature and have yet to be curated into open-access databases. To automate the process of extracting chemical structures from scientific papers, we developed the DECIMER.ai project. This open-source platform provides an integrated solution for identifying, segmenting, and recognising chemical structure depictions in scientific literature. DECIMER.ai comprises three main components: DECIMER-Segmentation, which utilises a Mask-RCNN model to detect and segment images of chemical structure depictions; DECIMER-Image Classifier EfficientNet-based classification model identifies which images contain chemical structures and DECIMER-Image Transformer which acts as an OCSR engine which combines an encoder-decoder model to convert the segmented chemical structure images into machine-readable formats, like the SMILES string.
DECIMER.ai is data-driven, relying solely on the training data to make accurate predictions without hand-coded rules or assumptions. The latest model was trained with 127 million structures and 483 million depictions (4 different per structure) on Google TPU-V4 VMs
Name to Structure Conversion: The conversion of structures to IUPAC(-like) or systematic names has been solved algorithmically or rule-based in satisfying ways. This fact, on the other side, provided us with an opportunity to generate a name-structure training pair at a very large scale to train a proof-of-concept transformer network and evaluate its performance.
In this work, the largest model was trained using almost one billion SMILES strings. The Lexichem software utility from OpenEye was employed to generate the IUPAC names used in the training process. STOUT V2 was trained on Google TPU-V4 VMs. The model's accuracy was validated through one-to-one string matching, BLEU scores, and Tanimoto similarity calculations. To further verify the model's reliability, every IUPAC name generated by STOUT V2 was analysed for accuracy and retranslated using OPSIN, a widely used open-source software for converting IUPAC names to SMILES. This additional validation step confirmed the high fidelity of STOUT V2's translations.