Article
Engineering, Manufacturing
Pengfei Guo, Moshe Haviv, Zhenwei Luo, Yulan Wang
Summary: This study investigates the server's best queue disclosure strategy in a single-server service system. The findings show that in a medium-sized market, the server's optimal strategy is often hybrid or mixed, involving randomizing queue concealment and revelation.
PRODUCTION AND OPERATIONS MANAGEMENT
(2022)
Article
Computer Science, Information Systems
Honggui Han, Jiacheng Zhang, Hongyan Yang, Ying Hou, Junfei Qiao
Summary: A data-driven robust optimal control (DROC) method is proposed to deal with uncertain nonlinear systems. The merits of the proposed method include a data-driven evaluation strategy, a multi-objective robust optimization algorithm, and the theoretical discussion on the robust boundedness of DROC. The effectiveness of DROC is demonstrated through experiments.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Theory & Methods
Jordi Garcia, Francesc Aguilo, Adria Asensio, Ester Simo, Marisa Zaragoza, Xavi Masip-Bruin
Summary: A new model is proposed for offloading task execution in heterogeneous environments, considering nodes computing capacity, network bandwidth, and geographical location. Two optimization strategies are suggested, and the simulation results show that the staged model provides the optimal solution in most scenarios.
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE
(2021)
Article
Management
Ruibing Wang, Qiao Wang, Wei-yu Kevin Chiang
Summary: The rapid development of digital technology has allowed firms to adopt innovative strategies for promoting their new products. This study investigates the efficient adoption of a promotional mix consisting of hype advertising campaign (HAC) and referral reward program (RRP). The research findings suggest that HAC and RRP should not always be adopted simultaneously, as consumer referrals may intensify wastage on HAC expenditure.
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH
(2024)
Article
Automation & Control Systems
Chase St. Laurent, Raghvendra Cowlagi
Summary: A coupled path-planning and sensor configuration method is proposed to minimize exposure to an unknown threat field. Gaussian Process regression is used to estimate the threat field from sensor measurements. The method introduces a task-driven information gain metric for sensor configuration and a surrogate metric for computational efficiency. It outperforms traditional decoupled information-driven sensor configuration in finding near-optimal plans.
Article
Computer Science, Interdisciplinary Applications
Huaming Tian, Yu Wang
Summary: A digital twin is created to continuously learn and improve model prediction in geotechnical projects using actual observation data. The challenges come from the spatial sparsity and spatiotemporal variations of the real geotechnical data. This study proposes a novel data-driven and physics-informed Bayesian learning framework to tackle these challenges and improve the model prediction.
COMPUTERS AND GEOTECHNICS
(2023)
Article
Mathematics, Applied
Duc-Lam Duong, Tapio Helin, Jose Rodrigo Rojo-Garcia
Summary: We study the stability properties of the expected utility function in Bayesian optimal experimental design. We provide a framework for this problem in a non-parametric setting and prove the convergence rate of the expected utility with respect to a likelihood perturbation. The assumptions set out for the general case are satisfied in the specific case of non-linear Bayesian inverse problems with Gaussian likelihood, and the stability of the expected utility with respect to perturbations to the observation map is regained. Numerical simulations are used to demonstrate the theoretical convergence rates in three different examples.
Article
Computer Science, Artificial Intelligence
Kaixin Lu, Zhi Liu, C. L. Philip Chen, Yaonan Wang, Yun Zhang
Summary: This article tackles the problem of removing the restriction on optimized performance caused by complicated design and heavy online parameter learning. It proposes a direct adaptive fuzzy inverse approach to design a switching-type inverse optimal controller and a one-parameter learning mechanism, ensuring input-to-state stability and achieving inverse optimality. The approach is verified through illustrative examples.
IEEE TRANSACTIONS ON FUZZY SYSTEMS
(2022)
Article
Environmental Sciences
Sara Nazif, Seyed Taghi Omid Naeeni, Zahra Akbari, Sara Fateri, Mohammad Ali Moallemi
Summary: This paper proposes a multilayer granular filter as an inexpensive and simple on-site treatment method for greywater reuse. An optimization-simulation model is developed to determine the best filter configuration. Experimental results show that a filter consisting of 33 cm of fine sand, 20 cm of activated carbon, and 7 cm of medium sand achieves the maximum efficiency of reducing COD and EC in greywater by 72% and 30% respectively.
JOURNAL OF ENVIRONMENTAL MANAGEMENT
(2023)
Article
Computer Science, Theory & Methods
Markus Hainy, David J. Price, Olivier Restif, Christopher Drovandi
Summary: A new approach using supervised classification methods is developed to perform Bayesian optimal model discrimination design, which requires fewer simulations and can assess performance through the misclassification error rate. This approach is particularly useful when dealing with models with intractable likelihoods and provides computational advantages when the likelihoods are manageable.
STATISTICS AND COMPUTING
(2022)
Article
Energy & Fuels
Mojeed Opeyemi Oyedeji, Abdullah Alharbi, Mujahed Aldhaifallah, Hegazy Rezk
Summary: In this study, data-driven models were developed using machine learning algorithms to optimize microbial fuel cells (MFCs), resulting in models with 99% accuracy on testing set evaluations. These models can be used to improve the power density and output voltage of MFCs.
Article
Automation & Control Systems
Longwen Liu, Wei Xie, Langwen Zhang
Summary: This article investigates the design technique of reduced-order interval observer (R-IO) for continuous-time linear systems with unknown external disturbances and measurement noises. The proposed coupled R-IO structure provides more design degrees of freedom to solve the difficulty of error system cooperativity construction and relax the constraint on sensor measurement noises. The existence condition of R-IO is derived as a set of matrix equations (MEs), and a complete solution is obtained by solving the MEs with available design parameters. An integrated optimization indicator is built to select these parameters. The efficiency of the obtained results is illustrated through numerical and practical examples.
INTERNATIONAL JOURNAL OF ROBUST AND NONLINEAR CONTROL
(2023)
Article
Automation & Control Systems
Shouping Guan, Jiashuo Li, Panpan Dong
Summary: This paper proposes an optimal interval controller design method for single input single output uncertain systems based on the Kharitonov theorem and an interval optimization algorithm. The optimal interval controller design is transformed into an optimal controller synthesis issue of multiple vertex objects using the Kharitonov theorem. An interval particle swarm optimization (IPSO) algorithm is used to obtain the solution domains of the controller parameters for each vertex object. The intersections of the solution domains for all vertex objects are obtained as the optimal interval solution of interval controller parameters.
ASIAN JOURNAL OF CONTROL
(2023)
Article
Engineering, Mechanical
Yichao Yang, Mayank Chadha, Zhen Hu, Michael D. Todd
Summary: This paper introduces a novel framework for optimal sensor placement design in structural health monitoring using Bayes risk as the objective function. The framework considers external and internal costs, making it applicable to various SHM designs. Through an example problem, the effectiveness and comprehensiveness of the framework are demonstrated, along with discussions on challenges such as computationally expensive models and uncertainty quantification.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2022)
Article
Ecology
Clinton B. Leach, Perry J. Williams, Joseph M. Eisaguirre, Jamie N. Womble, Michael R. Bower, Mevin B. Hooten
Summary: Optimal design procedures, when combined with Bayesian hierarchical models and recursive Bayesian computation, offer efficient tools for ecological learning and inference while reducing computational burdens. Examples using simulated data and real-world cases demonstrate the practical application of this method and emphasize the importance of computational gains for monitoring and science integration.
Article
Soil Science
Luciana Chavez Rodriguez, Ana Gonzalez-Nicolas, Brian Ingalls, Thilo Streck, Wolfgang Nowak, Sinan Xiao, Holger Pagel
Summary: This study applies prospective optimal design of experiments to identify laboratory sampling strategies that allow model-based discrimination of pesticide degradation pathways. The results highlight the importance of measuring pesticide metabolites for understanding pesticide fate in the environment. The study emphasizes the use of model-based prospective optimal design to maximize knowledge gains on soil systems from laboratory and field experiments.
EUROPEAN JOURNAL OF SOIL SCIENCE
(2022)
Article
Mechanics
Matthias Hinze, Sinan Xiao, Andre Schmidt, Wolfgang Nowak
Summary: This study evaluates and analyzes creep testing results on M2 salt concrete and determines the parameters of the fractional viscoelastic constitutive law using the Bayesian inversion method for reliable prediction of concrete behavior.
MECHANICS OF TIME-DEPENDENT MATERIALS
(2023)
Article
Environmental Sciences
Aline Schafer Rodrigues Silva, Tobias K. D. Weber, Sebastian Gayler, Anneli Guthke, Marvin Hoge, Wolfgang Nowak, Thilo Streck
Summary: There has been increasing interest in using multi-model ensembles, but a lack of methods to guide the choice of ensemble members. This study introduces a method based on energy statistics to quantify model similarities and assess goodness-of-fit. Visualization techniques are combined to support the interpretation of results. The case study on soil-plant-growth modeling demonstrates that model similarity and goodness-of-fit vary depending on the quantity of interest.
MODELING EARTH SYSTEMS AND ENVIRONMENT
(2022)
Article
Engineering, Aerospace
Sinan Xiao, Wolfgang Nowak
Summary: The reliability sensitivity index is a valuable tool to measure the impact of uncertain parameters on the failure of engineering systems. This study proposes a new efficient sampling method to estimate the failure-conditional PDF and reliability sensitivity index using a two-stage Markov chain Monte Carlo simulation. The method shows high efficiency compared to subset simulation and can handle high-dimensional problems.
AEROSPACE SCIENCE AND TECHNOLOGY
(2022)
Article
Energy & Fuels
Antonio Galvan, Jannik Haas, Simon Moreno-Leiva, Juan Carlos Osorio-Aravena, Wolfgang Nowak, Rodrigo Palma-Benke, Christian Breyer
Summary: While Europe and North America have conducted numerous studies on 100% renewable power systems, South America has lagged behind in research on zero-carbon energy systems. This study extends the existing energy planning model and applies it specifically to South America, resulting in a comprehensive model for the region's power sector. The study highlights the importance of solar photovoltaic, wind, gas, and concentrated solar power in South America's energy transition, and demonstrates that a shift to renewables is not only technically feasible but also cost-efficient. The study also explores the impact of spatial resolution and green hydrogen export scenarios on investment and cost. Overall, this research provides valuable insights for policymakers and the energy community in promoting sustainable transitions in South America.
Article
Computer Science, Interdisciplinary Applications
Ishani Banerjee, Peter Walter, Anneli Guthke, Kevin G. Mumford, Wolfgang Nowak
Summary: Bayesian model selection ranks competing models by computing Bayesian Model Evidence (BME) against test data. Computing BME can be problematic, and we propose a method called the Method of Forced Probabilities (MFP) to address this issue. We demonstrate the effectiveness of our approach on simulating gas migration models.
COMPUTATIONAL GEOSCIENCES
(2023)
Article
Environmental Sciences
Timothy Praditia, Matthias Karlbauer, Sebastian Otte, Sergey Oladyshkin, Martin V. V. Butz, Wolfgang Nowak
Summary: Improved understanding of complex hydrosystem processes is crucial for water resources research. Conventional modeling suffers from conceptual uncertainty, while machine learning models have limited generalization abilities. To address this, a hybrid modeling framework called FINN is proposed, which merges numerical methods with artificial neural networks. FINN yields interpretable results and its potential is demonstrated in a diffusion-sorption problem.
WATER RESOURCES RESEARCH
(2022)
Article
Computer Science, Interdisciplinary Applications
Rebecca Kohlhaas, Ilja Kroeker, Sergey Oladyshkin, Wolfgang Nowak
Summary: Surrogate models are widely used to improve computational efficiency in geophysical simulation problems. Existing multi-resolution PCE is a global representation that cannot estimate the uncertainty of the resulting surrogate. We propose combining multi-resolution PCE and GPE to correct surrogate bias and assess its uncertainty, resulting in a more stable emulator compared to GPE.
COMPUTATIONAL GEOSCIENCES
(2023)
Article
Environmental Sciences
Sebastian Schwindt, Sergio Callau Medrano, Kilian Mouris, Felix Beckers, Stefan Haun, Wolfgang Nowak, Silke Wieprecht, Sergey Oladyshkin
Summary: This study investigates the use of Bayesian calibration to identify faulty model setups and parameter combinations. Bayesian calibration utilizes a Gaussian process emulator as a surrogate model, which is faster than the actual numerical model. The results show that Bayesian calibration can describe the quality of calibration and correctness of model assumptions through geometric characteristics of posterior distributions.
WATER RESOURCES RESEARCH
(2023)
Editorial Material
Energy & Fuels
Liwei Zhang, Wolfgang Nowak, Sergey Oladyshkin, Yan Wang, Jianchao Cai
Summary: CO2 geological utilization and storage is an effective approach to reduce anthropogenic CO2 emissions. Recent advancements in modeling concepts, experimental approaches, safety assurance, and emerging technologies have driven the development of CO2 geological utilization and storage. A Sino-German joint symposium was organized to encourage global communication and collaboration in this field, bringing together experts from China, Germany, and other countries.
ADVANCES IN GEO-ENERGY RESEARCH
(2023)
Article
Environmental Sciences
Maria Fernanda Morales Oreamuno, Sergey Oladyshkin, Wolfgang Nowak
Summary: Bayesian model selection (BMS) and Bayesian model justifiability analysis (BMJ) provide a statistically rigorous framework for comparing competing models using Bayesian model evidence (BME). However, BME-based analysis has limitations in accounting for a model's predictive performance after calibration and in comparing models using different calibration subsets. To address these limitations, we propose augmenting BMS and BMJ analyses with information-theoretic measures such as expected log-predictive density (ELPD), relative entropy (RE), and information entropy (IE). We demonstrate how these measures, alongside BME, enhance the understanding of the Bayesian updating process and enable objective model comparison using different calibration datasets.
WATER RESOURCES RESEARCH
(2023)
Article
Computer Science, Information Systems
Christoph Dibak, Wolfgang Nowak, Frank Duerr, Kurt Rothermel
Summary: Numerical simulations on mobile devices are important for engineers and decision makers, but providing simulation results is challenging due to complexity and limited resources. This article presents an optimized approach using surrogate models and data assimilation to reduce communication overhead. Evaluation shows that the approach is 6.5 times faster than streaming from the server.
IEEE TRANSACTIONS ON MOBILE COMPUTING
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Matthias Karlbauer, Timothy Praditia, Sebastian Otte, Sergey Oladyshkin, Wolfgang Nowak, Martin V. Butz
Summary: The article introduces a physics-aware neural network called FINN, which combines the learning abilities of artificial neural networks with physical and structural knowledge from numerical simulation. The results demonstrate that FINN achieves higher modeling accuracy and excellent generalization ability, even with a significantly reduced number of parameters, outperforming other models.
INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 162
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Cosku Can Horuz, Matthias Karlbauer, Timothy Praditia, Martin V. Butz, Sergey Oladyshkin, Wolfgang Nowak, Sebastian Otte
Summary: Researchers have extended the finite volume neural network (FINN) to infer boundary condition values on-the-fly, demonstrating its ability to accurately infer unknown values, simulate Burgers' and Allen-Cahn equations, and generalize well beyond the encountered BC value range.
ARTIFICIAL NEURAL NETWORKS AND MACHINE LEARNING - ICANN 2022, PT I
(2022)
Article
Engineering, Multidisciplinary
Kai Cheng, Zhenzhou Lu, Sinan Xiao, Sergey Oladyshkin, Wolfgang Nowak
Summary: In this paper, a mixed covariance function Kriging model is proposed for uncertainty quantification. The model combines a traditional stationary covariance function and a nonstationary covariance function to represent the uncertainties. The optimal values of hyperparameters are obtained using a maximum likelihood estimation algorithm, and sparse representation is achieved by automatically removing small contribution basis functions. The model achieves comparable performance to the state-of-art models for nonlinear problems of moderate to high dimensionality.
INTERNATIONAL JOURNAL FOR UNCERTAINTY QUANTIFICATION
(2022)