Article
Engineering, Biomedical
Meng Zhang, Kevin B. Flores, Hien T. Tran
Summary: This study compared four data-driven models for closed-loop insulin delivery system development. Results showed that a neural network model had stable performance for short-term predictions, while regression models performed better at long-term prediction horizons and with lower computational costs.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2021)
Article
Engineering, Mechanical
Omid Bahrami, Wentao Wang, Rui Hou, Jerome P. Lynch
Summary: This study explores the use of data-driven models to forecast the response of bridges to truck loads. By deploying modern monitoring systems and utilizing advanced computer vision algorithms, the authors collected a unique dataset. The deep-learning models performed better in predicting bridge responses, reducing the forecasting error by at least 20% compared to traditional approaches.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2023)
Article
Environmental Sciences
Sheng Sheng, Kangling Lin, Yanlai Zhou, Hua Chen, Yuxuan Luo, Shenglian Guo, Chong -Yu Xu
Summary: Artificial neural networks have been increasingly used in water quality prediction due to their learning capability and generalizability. This study proposes a novel Encoder-Decoder model based on Temporal Convolutional Network for ammonia nitrogen forecasts. The results show that the developed TCN-ED model can accurately capture the complex relationships between ammonia nitrogen, water quality, and meteorological factors, outperforming other models in terms of accuracy and reliability.
JOURNAL OF ENVIRONMENTAL MANAGEMENT
(2023)
Article
Computer Science, Artificial Intelligence
Isidro Lloret, Jose A. Troyano, Fernando Enriquez, Juan-Jose Gonzalez-de-la-Rosa
Summary: Time series forecasting of disaggregated freight flow is a crucial issue for port authorities. This study tested two deep learning models on seven time series of imported goods from Morocco to Spain, achieving good results without preprocessing and highlighting the models' ability to work well with fixed input sizes for different granularities. The two neural network models both showed improvements in forecasting benchmarks, with the recurrent model performing best on daily data and the convolutional model excelling in weekly and monthly data.
Article
Thermodynamics
Huijuan Wu, Keqilao Meng, Daoerji Fan, Zhanqiang Zhang, Qing Liu
Summary: This paper proposes a multistep wind speed prediction model based on a transformer and shows through experiments that it achieves state-of-the-art performance in wind speed forecasting.
Article
Computer Science, Artificial Intelligence
Lakshika Girihagama, Muhammad Naveed Khaliq, Philippe Lamontagne, John Perdikaris, Rene Roy, Laxmi Sushama, Amin Elshorbagy
Summary: This study investigates the performance of sequence-to-sequence machine learning architectures in developing streamflow forecasting tools for Canadian watersheds. The attention-based encoder-decoder LSTM model outperformed the standard model in simulating overall hydrograph patterns. The results suggest that the attention mechanism in ML architectures is important and useful for hydrological applications, and the encoder-decoder LSTM with attention mechanism is a powerful choice for streamflow forecasting systems.
NEURAL COMPUTING & APPLICATIONS
(2022)
Article
Neurosciences
Tong Bai, Sen Zhou, Yu Pang, Jiasai Luo, Huiqian Wang, Ya Du
Summary: This paper proposes a novel method for addressing the challenges in image caption. The method introduces a guided decoding network to address visual information loss and non-dynamic adjustment of input images during decoding. The study adopts Dense Convolutional Network (DenseNet) and Multiple Instance Learning (MIL) in the image encoder, and Nested Long Short-Term Memory (NLSTM) as the decoder to enhance the extraction and parsing capability of image information. An attention mechanism and a double-layer decoding structure are incorporated to improve the model's performance in providing detailed descriptions and enriched semantic information. Deep Reinforcement Learning (DRL) is employed to train the model, optimizing evaluation indexes. The results demonstrate the model's improvement compared to commonly used models in evaluation indicators including BLEU, METEOR, and CIDEr.
FRONTIERS IN NEUROSCIENCE
(2023)
Article
Computer Science, Artificial Intelligence
Ziqiang Wang, Zhi Liu, Weijie Wei, Huizhan Duan
Summary: This paper introduces a deep convolutional neural network with a concise and effective architecture for saliency prediction, achieving top performance on two famous benchmarks. The model shows good generalization and performs better than other models, even when using the same backbone for comparison.
IMAGE AND VISION COMPUTING
(2021)
Article
Computer Science, Artificial Intelligence
Saban Gulcu
Summary: The training algorithm is a crucial component of artificial neural networks (ANN) that affects their performance. This article presents a new hybrid algorithm called DA-MLP, which uses the dragonfly algorithm to train feed-forward multilayer neural networks (MLP). The experimental study shows that the DA-MLP algorithm is more efficient than other algorithms.
APPLIED SOFT COMPUTING
(2022)
Article
Green & Sustainable Science & Technology
Ozge Cagcag Yolcu, Fulya Aydin Temel, Ayse Kuleyin
Summary: In this study, hybrid prediction models were used to estimate the adsorption of ammonium from landfill leachate by using zeolite in batch and column systems. The models showed significant improvement compared to traditional methods, with prediction errors being very low. The findings suggest that the proposed model can be effectively and reliably used without the need for additional experiments in environmental sciences.
JOURNAL OF CLEANER PRODUCTION
(2021)
Article
Thermodynamics
Hao Chen, Yngve Birkelund, Qixia Zhang
Summary: Accurate wind power forecasting is crucial in wind parks. Deep learning is increasingly used due to its ability to handle big data, but in-situ measured wind data are expensive and correlations between steps are often ignored. This study applies data augmentation to wind power forecasting and develops deep learning networks to improve forecasting accuracy.
ENERGY CONVERSION AND MANAGEMENT
(2021)
Article
Computer Science, Interdisciplinary Applications
Ahmed J. Aljaaf, Thakir M. Mohsin, Dhiya Al-Jumeily, Mohamed Alloghani
Summary: Using neural network models, the study accurately forecasted the outbreak of COVID-19 in Iraq, achieving accuracy rates of 87.6%, 82.4%, and 84.3% for daily infections, recovered cases, and deaths respectively. It is projected that by the end of September 2020, Iraq will have approximately 308,996 cases, with 228,551 expected to recover and 9,477 deaths.
JOURNAL OF BIOMEDICAL INFORMATICS
(2021)
Article
Energy & Fuels
Jakob Schmitt, Ivo Horstkoetter, Bernard Baeker
Summary: This study develops an encoder-decoder model framework based on recurrent neural networks, which is trained directly on unstructured battery data to simplify the modelling process. The proposed model shows excellent generalisation capabilities and time efficiency, making it suitable for applications such as online anomaly detection and power prediction.
JOURNAL OF ENERGY STORAGE
(2023)
Article
Computer Science, Artificial Intelligence
Shibai Yin, Yibin Wang, Yee-Hong Yang
Summary: A new attentive U-recurrent encoder-decoder dehazing network is proposed to address the three main limitations of convolutional neural networks in image dehazing, including ignoring relevant haze information, spatial inconsistency, and insufficient receptive field. The network combines an attentive recurrent network and a U-recurrent encoder-decoder network to improve spatial consistency, reduce information dilution, and enhance structural information capture. The experimental results demonstrate superior performance compared to state-of-the-art dehazing algorithms on both synthetic and real hazy images.
Article
Computer Science, Theory & Methods
Siping Liu, Xiaohan Tu, Cheng Xu, Renfa Li
Summary: Monocular depth estimation (MDE) plays a vital role in image sensing. However, current methods fail to provide satisfactory depth and are slow for inference on embedded devices. To tackle these problems, a novel encoder-decoder network (EDNet) is proposed.
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE
(2022)
Article
Engineering, Chemical
Sungil Yun, Matthew Tom, Junwei Luo, Gerassimos Orkoulas, Panagiotis D. Christofides
Summary: With increasing demands for microchips and the nano-scale semiconductor manufacturing industry, atomic layer etching (ALE) has become a critical etching process. This work develops microscopic models to characterize the thermal ALE process of aluminum oxide thin films with two precursors, setting the foundation for real-time, model-based operational parameter calculations.
CHEMICAL ENGINEERING RESEARCH & DESIGN
(2022)
Article
Engineering, Chemical
Mohammed S. Alhajeri, Fahim Abdullah, Zhe Wu, Panagiotis D. Christofides
Summary: This work investigates the impact of Gaussian and non-Gaussian noise on the performance of partially-connected recurrent neural network models used for modeling chemical processes. Two techniques, Monte Carlo dropout and co-teaching, are employed to reduce overfitting and improve model accuracy and controller performance.
CHEMICAL ENGINEERING RESEARCH & DESIGN
(2022)
Article
Engineering, Chemical
Mohammed S. Alhajeri, Aisha Alnajdi, Fahim Abdullah, Panagiotis D. Christofides
Summary: To approximate nonlinear dynamic systems using time-series data, recurrent neural networks (RNNs) and long short-term memory (LSTM) networks are often used. Incorporating prior knowledge in machine learning-based models can further improve accuracy. This study develops a methodological framework to quantify the generalization error bounds for partially-connected RNNs and LSTM models. The proposed approach shows improved performance in predictive control systems compared to traditional approaches under Lyapunov-based MPC.
CHEMICAL ENGINEERING RESEARCH & DESIGN
(2023)
Article
Engineering, Chemical
Fahim Abdullah, Mohammed S. Alhajeri, Panagiotis D. Christofides
Summary: SINDy is a nonlinear modeling technique that shows superior performance in handling time-series data but requires careful consideration of noise. In this study, SINDy is combined with ensemble learning to identify multiple models for improving overall nonlinear model performance.
INDUSTRIAL & ENGINEERING CHEMISTRY RESEARCH
(2022)
Article
Engineering, Chemical
Atharva Suryavanshi, Aisha Alnajdi, Mohammed Alhajeri, Fahim Abdullah, Panagiotis D. Christofides
Summary: In this work, secure and private communication links are established between sensor-controller and controller-actuator elements using semi-homomorphic encryption to ensure cyber-security in model predictive control (MPC) of nonlinear systems. The Paillier cryptosystem is implemented for encryption-decryption operations in the communication links. The closed-loop encrypted MPC is designed with a certain degree of robustness to the quantization errors in nonlinear systems, and the trade-off between accuracy and computational cost is discussed. Chemical process examples are employed to demonstrate the implementation of the proposed encrypted MPC design.
Article
Engineering, Chemical
Berkay Citmaci, Junwei Luo, Joon Baek Jang, Carlos G. Morales-Guio, Panagiotis D. Christofides
Summary: The electrochemical reduction of CO2 gas is a new technique for mitigating the global climate crisis and storing energy from renewable sources. However, there is a lack of explicit models for CO2 reduction and limited effort in developing process modeling and control of CO2 electrochemical reactors. This study focuses on developing a control scheme for a rotating cylinder electrode (RCE) reactor using artificial and recurrent neural network modeling, nonlinear optimization, and process controller design. The experimental results demonstrate the effectiveness of the control system in regulating the production rates of ethylene and carbon monoxide.
CHEMICAL ENGINEERING RESEARCH & DESIGN
(2023)
Article
Computer Science, Interdisciplinary Applications
Fahim Abdullah, Panagiotis D. Christofides
Summary: This paper discusses recent developments in the data-based modeling and control of nonlinear chemical process systems using sparse identification of nonlinear dynamics (SINDy). Challenges of handling time-scale multiplicities and noisy sensor data when using SINDy are addressed, and novel methods devised to overcome these challenges are described. Modeling guidelines for using the proposed techniques for process systems are provided.
COMPUTERS & CHEMICAL ENGINEERING
(2023)
Article
Engineering, Chemical
Fahim Abdullah, Panagiotis D. Christofides
Summary: This study presents a sparse identification-based model predictive control framework that incorporates online updates of the sparse-identified model to consider nonlinear dynamics and model uncertainty in process systems. The method involves obtaining a nonlinear first-order ordinary differential equation model using sparse identification for nonlinear dynamics (SINDy), which is integrated into Lyapunov-based MPC (LMPC) and Lyapunov-based economic MPC (LEMPC) for steady-state operation and optimal economic performance. An online model update scheme is proposed to improve prediction accuracy, utilizing prediction errors and process data. The proposed methodology is demonstrated to enhance dynamic performance and ensure closed-loop stability and optimality in a chemical process example.
CHEMICAL ENGINEERING RESEARCH & DESIGN
(2023)
Article
Engineering, Chemical
Yash A. Kadakia, Atharva Suryavanshi, Aisha Alnajdi, Fahim Abdullah, Panagiotis D. Christofides
Summary: This work focuses on developing and applying Encrypted Lyapunov-based Model Predictive Control (LMPC) in a nonlinear chemical process network for Ethylbenzene production. The Paillier cryptosystem is employed for encryption-decryption operations in the communication channels between the sensor-controller and controller-actuator to enhance system cybersecurity. The impact of quantization parameter on the performance of the controller and the overall encryption to control input calculation time is examined through closed-loop simulations under the encrypted LMPC scheme.
Article
Mathematics
Aisha Alnajdi, Fahim Abdullah, Atharva Suryavanshi, Panagiotis D. Christofides
Summary: In this study, a general form of nonlinear two-time-scale systems is presented and machine learning techniques are used to approximate the dynamics of both subsystems. A Lyapunov-based model predictive control scheme is designed, and conditions for closed-loop stability are derived. The theory is validated with a nonlinear chemical process example.
Article
Engineering, Chemical
Yash A. Kadakia, Aisha Alnajdi, Fahim Abdullah, Panagiotis D. Christofides
Summary: This research focuses on encrypted distributed control architectures to improve the operational safety, cybersecurity, and computational efficiency of large-scale nonlinear systems. The study demonstrates the effectiveness of using encrypted DMPC with state estimation in scenarios where only partial state measurements are available.
DIGITAL CHEMICAL ENGINEERING
(2023)
Article
Engineering, Chemical
Yash A. Kadakia, Aisha Alnajdi, Fahim Abdullah, Panagiotis D. Christofides
Summary: This work aims to enhance the operational safety, cybersecurity, computational efficiency, and closed-loop performance of large-scale nonlinear time-delay systems by employing a decentralized model predictive controller with encrypted networked communication. The nonlinear process is partitioned into multiple subsystems, each controlled by a distinct Lyapunov-based MPC. Predictors are integrated to handle input and state delays, while encryption is used to enhance cybersecurity.
CHEMICAL ENGINEERING RESEARCH & DESIGN
(2023)
Article
Computer Science, Interdisciplinary Applications
Yash A. Kadakia, Atharva Suryavanshi, Aisha Alnajdi, Fahim Abdullah, Panagiotis D. Christofides
Summary: This study presents an encrypted two-tier control architecture integrated with a machine learning-based cyberattack detector to enhance the safety, security, and performance of nonlinear processes. The upper tier utilizes an encrypted nonlinear model predictive controller to improve performance, while the lower tier uses an encrypted set of linear controllers for stabilization. To mitigate the risk of cyberattacks, a machine learning-based detector is developed to identify attacks and switch to the secure lower tier. The study includes a comprehensive stability analysis and provides practical implementation guidelines.
COMPUTERS & CHEMICAL ENGINEERING
(2024)
Article
Engineering, Chemical
Aisha Alnajdi, Atharva Suryavanshi, Mohammed S. Alhajeri, Fahim Abdullah, Panagiotis D. Christofides
Summary: The purpose of this work is to study machine-learning-based model predictive control of nonlinear systems with time-delays. The proposed approach involves initially building a machine learning model (i.e., Long Short Term Memory (LSTM)) to capture the process dynamics in the absence of time delays. Then, an LSTM-based model predictive controller (MPC) is designed to stabilize the nonlinear system without time delays. Closed-loop stability results are then presented, establishing robustness of this LSTM-based MPC towards small time-delays in the states. To handle input delays, we design an LSTM-based MPC with an LSTM-based predictor that compensates for the effect of input delays. The predictor is used to predict future states using the process measurement, and then the predicted states are used to initialize the LSTM-based MPC. Stabilization of the time-delay system with both state and input delays around the steady state is achieved through the featured design. The approach is applied to a chemical process example, and its performance and robustness properties are evaluated via simulations.
DIGITAL CHEMICAL ENGINEERING
(2023)
Article
Computer Science, Interdisciplinary Applications
Nohan Joemon, Melpakkam Pradeep, Lokesh K. Rajulapati, Raghunathan Rengaswamy
Summary: This paper introduces a smoothing-based approach for discovering partial differential equations from noisy measurements. The method is data-driven and improves performance by incorporating first principles knowledge. The effectiveness of the algorithm is demonstrated in a real system using a new benchmark metric.
COMPUTERS & CHEMICAL ENGINEERING
(2024)
Article
Computer Science, Interdisciplinary Applications
Zhibin Lu, Yimeng Li, Chang He, Jingzheng Ren, Haoshui Yu, Bingjian Zhang, Qinglin Chen
Summary: This study proposes a new inverse design method using a physics-informed neural network to identify optimal heat sink designs. A hybrid PINN accurately approximates the governing equations of heat transfer processes, and a surrogate model is constructed for integration with optimization algorithms. The proposed method accelerates the search for Pareto-optimal designs and reduces search time. Comparing different scenarios facilitates real-time observation of multiphysics field changes, improving understanding of optimal designs.
COMPUTERS & CHEMICAL ENGINEERING
(2024)
Article
Computer Science, Interdisciplinary Applications
Luca Gasparini, Antonio Benedetti, Giulia Marchese, Connor Gallagher, Pierantonio Facco, Massimiliano Barolo
Summary: In this paper, a method for batch process monitoring with limited historical data is investigated. The methodology utilizes machine learning algorithms to generate virtual data and combines it with real data to build a process monitoring model. Automatic procedures are developed to optimize parameters, and indicators and metrics are proposed to assist virtual data generation activities.
COMPUTERS & CHEMICAL ENGINEERING
(2024)
Article
Computer Science, Interdisciplinary Applications
Julia Jimenez-Romero, Adisa Azapagic, Robin Smith
Summary: Energy transition is a significant and complex challenge for the industry, and developing cost-effective solutions for synthesizing utility systems is crucial. The research combines mathematical formulation with realistic configurations and conditions to represent utility systems and provides a basis for synthesizing energy-efficient utility systems for the future.
COMPUTERS & CHEMICAL ENGINEERING
(2024)
Article
Computer Science, Interdisciplinary Applications
Samuel Adeyemo, Debangsu Bhattacharyya
Summary: This work develops algorithms for estimating sparse interpretable data-driven models. The algorithms select the optimal basis functions and estimate the model parameters using Bayesian inferencing. The algorithms estimate the noise characteristics and model parameters simultaneously. The algorithms also exploit prior analysis and special properties for efficient pruning, and use a modified Akaike information criterion for model selection.
COMPUTERS & CHEMICAL ENGINEERING
(2024)
Article
Computer Science, Interdisciplinary Applications
Abbasali Jafari-Nodoushan, Mohammad Hossein Dehghani Sadrabadi, Maryam Nili, Ahmad Makui, Rouzbeh Ghousi
Summary: This study presents a three-objective model to design a forward supply chain network considering interrelated operational and disruptive risks. Several strategies are implemented to cope with these risks, and a joint pricing strategy is used to enhance the profitability of the supply chain. The results show that managing risks and uncertainties simultaneously can improve sustainability goals and reduce associated costs.
COMPUTERS & CHEMICAL ENGINEERING
(2024)
Article
Computer Science, Interdisciplinary Applications
T. A. Espaas, V. S. Vassiliadis
Summary: This paper extends the concept of higher-order search directions in interior point methods to convex nonlinear programming. It provides the mathematical framework for computing higher-order derivatives and highlights simplified computation for special cases. The paper also introduces a dimensional lifting procedure for transforming general nonlinear problems into more efficient forms and describes the algorithmic development required to employ these higher-order search directions.
COMPUTERS & CHEMICAL ENGINEERING
(2024)
Article
Computer Science, Interdisciplinary Applications
David A. Linan, Gabriel Contreras-Zarazua, Eduardo Sanhez-Ramirez, Juan Gabriel Segovia-Hernandez, Luis A. Ricardez-Sandoval
Summary: This study proposes a parallel hybrid algorithm for optimal design of process flowsheets, which combines stochastic method with deterministic algorithm to achieve faster and improved convergence.
COMPUTERS & CHEMICAL ENGINEERING
(2024)
Article
Computer Science, Interdisciplinary Applications
Xiaoyong Lin, Zihui Li, Yongming Han, Zhiwei Chen, Zhiqiang Geng
Summary: A novel GAT-LSTM model is proposed for the production prediction and energy structure optimization of propylene production processes. It outperforms other models and can provide the optimal raw material scheme for actual production processes.
COMPUTERS & CHEMICAL ENGINEERING
(2024)
Article
Computer Science, Interdisciplinary Applications
Prodromos Daoutidis, Jay H. Lee, Srinivas Rangarajan, Leo Chiang, Bhushan Gopaluni, Artur M. Schweidtmann, Iiro Harjunkoski, Mehmet Mercangoz, Ali Mesbah, Fani Boukouvala, Fernando Lima, Antonio del Rio Chanona, Christos Georgakis
Summary: This paper provides a concise perspective on the potential of machine learning in the PSE domain, based on discussions and talks during the FIPSE 5 conference. It highlights the need for domain-specific techniques in molecular/material design, data analytics, optimization, and control.
COMPUTERS & CHEMICAL ENGINEERING
(2024)
Article
Computer Science, Interdisciplinary Applications
Hesam Hassanpour, Prashant Mhaskar, Brandon Corbett
Summary: This work addresses the problem of designing an offset-free implementable reinforcement learning (RL) controller for nonlinear processes. A pre-training strategy is proposed to provide a secure platform for online implementations of the RL controller. The efficacy of the proposed approach is demonstrated through simulations on a chemical reactor example.
COMPUTERS & CHEMICAL ENGINEERING
(2024)
Article
Computer Science, Interdisciplinary Applications
Hunggi Lee, Donghyeon Lee, Jaewook Lee, Dongil Shin
Summary: This study introduces an innovative framework that utilizes a limited number of sensors to detect chemical leaks early, mitigating the risk of major industrial disasters, and providing faster and higher-resolution results.
COMPUTERS & CHEMICAL ENGINEERING
(2024)
Article
Computer Science, Interdisciplinary Applications
Sibel Uygun Batgi, Ibrahim Dincer
Summary: This study examines the environmental impacts of three alternative hydrogen-generating processes and determines the best environmentally friendly option for hydrogen production by comparing different impact categories. The results show that the solar-based HyS cycle options perform the best in terms of global warming potential, abiotic depletion, acidification potential, ozone layer depletion, and human toxicity potential.
COMPUTERS & CHEMICAL ENGINEERING
(2024)
Article
Computer Science, Interdisciplinary Applications
LaGrande Gunnell, Bethany Nicholson, John D. Hedengren
Summary: A review of current trends in scientific computing shows a shift towards open-source and higher-level programming languages like Python, with increasing career opportunities in the next decade. Open-source modeling tools contribute to innovation in equation-based and data-driven applications, and the integration of data-driven and principles-based tools is emerging. New compute hardware, productivity software, and training resources have the potential to significantly accelerate progress, but long-term support mechanisms are still necessary.
COMPUTERS & CHEMICAL ENGINEERING
(2024)
Article
Computer Science, Interdisciplinary Applications
Daniel Cristiu, Federico d'Amore, Fabrizio Bezzo
Summary: This study presents a multi-objective mixed integer linear programming framework to optimize the supply chain for mixed plastic waste in Northern Italy. Results offer quantitative insights into economic and environmental performance, balancing trade-offs between maximizing gross profit and minimizing greenhouse gas emissions.
COMPUTERS & CHEMICAL ENGINEERING
(2024)