67639 research outputs found
Sort by
Control and estimation designs using model-based and model-free approaches for water quality monitoring in process systems
No full textThis dissertation addresses the design and evaluation of advanced control and estimation approaches in process systems, with a particular emphasis on water-related application systems such as aquaculture systems and wastewater treatment plants (WWTPs). Despite the differences in the environmental operations in aquaculture and WWTP systems, they both share challenges, such as complex system dynamics, limited direct measurements, and the need for robust monitoring and decision-making strategies.
In aquaculture, the ultimate objective is to promote optimal fish growth while minimizing food waste. Various control strategies—including bang-bang control, proportional-integral-derivative (PID) control, model predictive control (MPC), and reinforcement learning (RL) via Q-learning—were investigated under various un-ionized ammonia (UIA), a crucial parameter influencing fish mortality. Each controller effectively tracked the desired fish growth trajectory; however, MPC demonstrated superior efficiency by reducing food consumption due to its ability to incorporate constraints and multiple-input objectives. Conversely, reinforcement learning (RL) control provided a feeding policy and maintained a relative food consumption that might underfeed the fish. To address the absence of a direct commercial fish weight sensor, this dissertation introduces an extended Kalman filter (EKF) that effectively estimates fish weight under varying initial conditions and noises.
For WWTPs, this dissertation proposes data-driven approaches for bacterial concentration estimation, a key indicator of treatment effectiveness. Given the challenges of limited data and lack of model generalizability across plants, decision tree models—such as k-nearest neighbors, random forest, gradient boosting regression, and extreme gradient boosting—were evaluated for bacterial concentration estimation. Additionally, a calibration framework was proposed to enhance the generalization of neural network (NN) models by continuously updating the trained model with receiving new sample
On the flow characteristics in the shock formation region due to the diaphragm opening process in a shock tube
No full textThe shock formation process in shock tubes has been extensively studied; however, the influence of diaphragm rupture dynamics on the resulting flow non-uniformities remains inadequately understood. Existing models predicting the shock attenuation and propagation dynamics overlook critical diaphragm mechanics and their impact on shock behavior. Addressing this gap is vital for improving predictive capabilities and optimizing shock tube designs for applications in combustion kinetics, aerodynamics, and high-speed diagnostics. This study investigates the shock wave formation through combined experimental and numerical approaches over a range of driver-to-driven pressure ratios (Driver pressure: 9.4–25.5 bar of helium; Driven pressure: 100 Torr (133.322 mbar) of argon). High-speed imaging is used to capture the diaphragm opening dynamics, while pressure and shock velocity measurements along the entire driven section of the shock tube provide key validation data for computational fluid dynamic simulations. Two-dimensional numerical simulations incorporate experimentally measured diaphragm opening profiles, offering detailed insights into flow features and thermodynamic gradients behind the moving shock front. Key parameters, including deceleration and acceleration phases within the shock formation region, shock formation distances, and times, have been quantified. A novel theoretical framework is introduced to correlate these parameters, enabling accurate predictions of shock Mach number evolution under varying conditions. This unified methodology bridges theoretical and experimental gaps, providing a robust foundation for advancing shock tube research and design.This work was sponsored by King Abdullah University of Science and Technology (KAUST) and supported by the KAUST Supercomputing Laboratory (KSL). All simulations were performed on KSL's Shaheen III supercomputer. Convergent Science provided CONVERGE licenses and technical support for this work. The authors appreciate Md Zafar Ali Khan's assistance in the shock tube experiments
Electrochemical Platform for Monitoring Neurotransmitters and Therapeutic Drugs for Neural Diseases
No full textNeurodegenerative diseases (NDs) are a heterogeneous group of disorders characterized by the progressive dysfunction and degeneration of neurons within the central nervous system (CNS). Prominent examples include Alzheimer’s disease (AD), Parkinson’s disease (PD), Huntington’s disease, and amyotrophic lateral sclerosis (ALS). Among the multiple factors contributing to their onset, ageing is widely recognized as the most significant risk factor. Early detection and precise monitoring of neurotransmitters and neuroactive drugs are therefore essential for understanding disease progression and for improving therapeutic strategies.
In this thesis, a series of electrochemical sensing platforms were developed for monitoring neurotransmitters and neural drugs, focusing on dopamine, L-DOPA, and vancomycin. First, a laser-scribed graphene (LSG) electrode decorated with a metal organic framework (MOF) based material was fabricated for the detection of dopamine using both electrochemical and colorimetric approaches. Building upon this, a multiplexed LSG electrode decorated with single-atom catalysts was developed and successfully applied for sensitive dopamine detection. The platform was then extended to microneedle-based electrodes, where LSG microneedles decorated with dual single-atom catalysts enabled the minimally invasive detection of L-DOPA directly in interstitial fluid (ISF). Finally an aptamer functionalized microneedle array was engineered for the continuous monitoring of vancomycin, a clinically relevant narrow therapeutic index antibiotic. This platform was validated in vivo in a mouse model, demonstrating its potential for real-time therapeutic drug monitoring.
Collectively, these results highlight the versatility of advanced electrochemical biosensing platforms ranging from MOF decorated graphene to single atom catalyst modified microneedles for neurotransmitter and drug monitoring. The technologies developed in this chapter not only offer new opportunities for personalized medicine in neurodegenerative diseases but also provide a foundation for future minimally invasive, real-time sensing strategies in clinical settings
Computational and Statistical Advances in Spatio-Temporal Modeling: Causality, Deep Learning, and High-Performance Computing
No full textRecent advances in environmental monitoring and remote sensing have led to an unprecedented increase in spatial and spatio-temporal data complexity, presenting both opportunities and challenges for environmental science. This thesis explores three critical challenges in environmental data analysis. First, we investigate spatial causality in spatio-temporal systems, proposing new methodologies to handle hidden confounding variables, extreme events, and structural nonlinearities. Second, we examine spatio-temporal forecasting for wind energy, exploring the application of advanced deep learning architectures, including deep echo state networks and graph autoencoders to capture complex nonlinear wind patterns. Third, we address computational limitations of a special mathematical function, \textsc{BesselK}, that is widely used in Gaussian processes with Mat\'{e}rn kernel, developing GPU-accelerated matrix generation algorithms for efficient processing of large-scale spatial datasets, particularly focusing on maximum likelihood estimation. Through comprehensive empirical validation using simulated and real-world datasets, we demonstrate the effectiveness of our proposed approaches. Our work contributes to environmental data science by addressing fundamental computational and methodological challenges in analyzing large-scale spatial and spatio-temporal datasets
Analysis of Deformation Characteristics of Layered Rock Tunnel Excavation Based on Statistical Mechanics of Rock Masses
No full textLayered rock mass is prevalent in tunnel engineering, where deformation and failure of surrounding rock are primarily governed by discontinuity structural planes, leading to significant anisotropy in deformation and strength characteristics. Compared to continuous homogeneous rock masses, layered rock masses exhibit more complex engineering properties. This study utilized the constitutive theory of statistical mechanics of rock masses (SMRM) and Abaqus numerical software to analyze the influence of geometric, spatial, and mechanical characteristics of discontinuity structural planes on the stability of surrounding rock. The failure characteristics of tunnel surrounding rock with varying scales and orientations of carbonaceous slate discontinuity structural planes are examined. Additionally, the displacement distribution of surrounding rock under continuous medium conditions is compared between the SMRM constitutive theory and the Mohr–Coulomb constitutive theory. The results reveal that the SMRM constitutive theory aligns well with the Mohr–Coulomb theory under continuous medium conditions. The presence of discontinuity structural planes significantly diminishes the self-stability of the surrounding rock, with increased scale and density of structural planes amplifying their control effect. The direction of surrounding rock failure postexcavation is closely related to the formation occurrence and horizontal stress, manifesting as tangential shear sliding failure along the discontinuity structural plane and normal bending compression shear failure of the bedrock. Meanwhile, the model test results are employed to validate the feasibility of SMRM theory. This study provides an important reference for the analysis of deformation and failure mechanisms of layered rock masses with different joint surface parameters.The authors gratefully acknowledge the support of the National Natural Science Foundation of China (Grant No. 42207199), Zhejiang Province International Science and Technology Cooperation Base Open Fund Project (IBGDP-2023-01), and Zhejiang Provincial Postdoctoral Science Foundation (Grant No. ZJ2022156)
Expanded Synthesis of 3D Covalent Organic Frameworks via Linker Exchange for Efficient Photocatalytic Aerobic Oxidation.
No full textDespite recent progress in 3D covalent organic frameworks (3D-COFs), their design and synthesis still pose significant challenges, mainly due to a limited mechanistic understanding of their synthesis. Herein, a linker exchange approach has been utilized to synthesize a series of new 3D-COFs by first preparing an imine-linked 3D-COF followed by exchanging with selected linear diamine linkers. This approach can be widely applicable to different types of diamines, enabling rational-designed synthesis of 3D frameworks that are previously inaccessible via direct polymerization in a one-pot reaction. Mechanistic aspects associated with the improved 3D-COF synthesis via the linker exchange approach, are investigated by density functional theory calculations, in which the possibility of the departure of the leaving linker is a spontaneous process with a decrease in enthalpy. Catalytic and computational results revealed that incorporating benzoxazole moiety into the 3D-COF frameworks enables a significant increase in the capability of visible-light-driven catalysis. The overall findings of the present study will pave the way toward the development of 3D-COFs with tunable structures and functions for other promising and challenging applications.W.Z. and Z.L. contributed equally to this work. The authors acknowledgethe Robert A. Welch Foundation (B-0027) for financial support of this work.Partial support from the Researchers Supporting Program (RSP2025R55)at King Saud University, Riyadh, Saudi Arabia (AMA) is also acknowledged
Interface Engineering Strategies for Efficient Perovskite and Organic Light-Emitting Diodes
No full textOver the past years, significant academic progress has been made in enhancing the performance of perovskite light-emitting diodes and organic light-emitting diodes. However, the market demand for devices with even higher efficiencies and improved stability remains pressing. At the same time, the global energy crisis underscores the need to reduce the energy consumption of light-emitting diodes. In this context, understanding and optimizing the interface of (nano)materials within these devices is essential for further performance advancement. This thesis demonstrates interface engineering approaches that not only increase device efficiency but also simplify device complexity and reduce costs. The proposed methods are straightforward, highly versatile, and applicable to advancing optoelectronics.
In the first study, we present efficient green perovskite-based light-emitting diodes utilizing perovskite: small molecule blends. We find that this molecule improves the perovskite morphology. By combining the molecular additive with hole-injecting self-assembled monolayers, we achieve devices of significantly enhanced efficiency and stability, reaching a quantum efficiency of up to ~19%. This work underscores the potential of perovskite: small molecule blends as a promising method for achieving efficient perovskite light-emitting diodes, while the usage of self-assembled monolayers enables the development of simpler and more reliable materials for application in next-generation light-emitting diodes.
In the second study, we conduct a comprehensive study of various self-assembled monolayers to improve hole-injection through the surface modification of indium tin oxide anode, a straightforward method for enhancing device efficiency. The physicochemical properties of the modified anodes are characterized using a set of various techniques. To investigate the hole-injection properties of these different
5
molecules, we fabricate and characterize green phosphorescent organic light-emitting devices. Our findings highlight that a particular molecule yields devices with superior performance characteristics, including an efficiency of up to ~17%. This improved performance can be attributed to several synergistic factors, including the deep work function of this molecule, its ability to form large molecular clusters, its moderate conductivity, and the presence of intrinsic dipoles, all contributing to the enhancement of hole-injection.
In the third study, we demonstrate the potential of an optimized two-step post-treatment of poly(3,4-ethylenedioxythiophene) polystyrene sulfonate, which increases the electrical conductivity up to 5,900 S cm-1. Post-treated samples are employed as anodes in the fabrication of both organic and perovskite-based devices, achieving efficiencies of ~14% and ~10%, respectively. Additionally, we demonstrate that these anodes can be deposited on flexible substrates, enabling functional flexible devices. The results of this work highlight that the suggested post-treatment method produces highly conductive films, which are significantly promising and versatile transparent electrode materials for cost-effective and flexible optoelectronic devices
Organic semiconductor detectors for alpha and neutron detection
No full textOrganic semiconductor (OSC) technologies are of active scientific interest due to their tunable, scalable, and cost-effective nature. We will present radiation sensors based on OSC technology, particularly applications related to detection of hadronic radiation consisting of
radiation and thermal and fast neutrons. Neutron detection is useful in various fields, from fundamental particle and atomic physics research to the medical field and nuclear security portal monitors. These organic sensors focus on NDI-type organic polymers including a novel material with carborane, a polyhedral cluster of carbon, boron, and hydrogen, directly incorporated in the molecular backbone (oCbT2-NDI), sensitising to thermal neutrons via the boron neutron capture (BNC) process. A comparison is made with a similar polymer (PNDI(2OD)2T) with homogeneously dispersed boron carbide (B4C) microparticles, and a control sensor without any boron which is sensitive to more energetic fast neutrons.The authors gratefully acknowledge the generous contributions of the following UK organisations: AWE Plc. (UK Ministry of Defence © Crown owned copyright 2025/AWE), Queen Mary University of London, the UK Home Office, the Nuclear Security Science Network, and the Science and Technology Facilities Council, United Kingdom to support this work. Furthermore we acknowledge the staff at NPL as well as the Technical Staff within the School of Physical and Chemical Sciences at Queen Mary University of London for technical support and assistance in completing this work
Selective Electrocatalytic CO <sub>2</sub> Reduction to Methanol: A Roadmap toward Practical Implementation
No full textElectrocatalytic CO2 reduction to methanol (MeOH) unites two urgent global needs, carbon recycling and renewable energy storage, into a single, compelling chemical transformation. According to recent techno-economic analyses, commercially competitive MeOH production (at ≈$190 per ton) can be achieved via electroreduction by meeting practical targets for current density, Faradaic efficiency (FE), and stability. Moreover, MeOH's high energy density (16 MJ L−1), substantial hydrogen content (100 g H2 per L), and low storage and transport costs further underscore its strong economic potential. Yet, the complexity of the six-electron–proton transfer (ET–PT) process that governs its formation remains intrinsically complex, with competing pathways threatening selectivity at every stage. This review critically examines current mechanistic insights, highlighting key intermediates such as CO and OCH3, and demonstrating how catalyst surfaces and reaction conditions profoundly influence pathway divergence. We highlight recent advances in catalyst development that exploit a fundamental, molecular-level understanding of intermediate stabilization to deliver unprecedented MeOH selectivity and activity. Through detailed analysis of operational parameters—including mass transport dynamics, electrolyte composition, and applied potentials—this work provides a comprehensive framework for rational catalyst development. Together, these insights converge design principles for next-generation electrocatalysts capable of selectively converting CO2-to-MeOH at scale, advancing economically viable and environmentally sustainable MeOH production
Who invented deep residual learning?
No full textModern AI is based on deep artificial neural networks (NNs). As of 2025, the most cited scientific article of the 21st century is an NN paper on deep residual learning with residual connections. Who invented this? We present a timeline of the evolution of deep residual learning.Thank you to Lucas Beyer, Francois Chollet, Kazuki Irie, Rupesh Kumar Srivastava, Sepp Hochreiter, Felix Gers, and others, for their helpful feedback. (Let me know under [email protected] if you can spot any remaining error.) The contents of thi