Sustainable Development toward Low-Carbon Energy and Industry Future: Transferable Cross-Scale Eco-chemical Industry Parks

Energy & Fuels, Journal Year: 2025, Volume and Issue: 39(6), P. 3375 - 3382

Published: Jan. 31, 2025

Industrial decarbonization is a global challenge requiring collective efforts, with the chemical industry, as significant emitter, bearing substantial responsibility. The introduction of eco-industrial park concept aims to link factories within park, integrating operations at different levels achieve overall optimization and provide solutions for carbon reduction in industry. This study first explores direction industry recommends development standardized management framework parks. To address these research gaps, transferable cross-scale proposed. optimizes coordination internal external production conditions through scheduling provides targeted evaluation system. surrogate models enhance flexibility transferability framework. Overall, offers modeling approach adaptable multiscale characteristics which aim optimize strategies.

Language: Английский

---

Calibration-Augmented and Mechanism-Driven Deep Learning Hybrid Framework for Modeling Actual Distillation Processes

Industrial & Engineering Chemistry Research, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 6, 2025

Accurate prediction models are pivotal to improving the production efficiency and ensuring product quality in distillation processes. Traditional mechanism-based neglect real-world fluctuations, while data-driven suffer from noise overlook chemical constraints, leading inaccurate data diminished performance. Therefore, a hybrid framework that embeds model calibration into deep learning is proposed leverage complementary capabilities of both methodologies. The solves problem insufficient accuracy by calibration, including nonparameter regression, liquid level correction, robust estimator. It also takes thermodynamic constraints account integrating with convolutional neural network (CNN), thereby capturing dynamic relationships between variables efficiently predicting key process parameters. calibration-augmented mechanism-driven CNN achieves exceptional predictive performance, validating effectiveness complex modeling, further offering novel insight paradigms for digital twin intelligent factories.

Language: Английский

---

Short‐Term Target Maneuvering Trajectory Prediction Using DTW–CNN–LSTM

International Journal of Aerospace Engineering, Journal Year: 2025, Volume and Issue: 2025(1)

Published: Jan. 1, 2025

This paper introduces a novel prediction model designed to mitigate the substantial data dependency associated with maneuver trajectory in unmanned combat air vehicles (UCAVs) during combat. Considering characteristics of high noise, dynamic complexity, and variable lengths inherent short‐range scenarios, we employ time warping (DTW) assess similarity 3D data. approach allows us identify select most analogous historical data, which then utilize as our training dataset. In pursuit enhanced precision for online prediction, propose an improved convolutional neural network (CNN) that not only offers “after‐zero” information but also incorporates delay compensation mechanisms. Our experimental findings indicate proposed satisfies stringent timeliness requirements outperforms benchmark models terms accuracy across various operating conditions.

Language: Английский

---

Prediction and multi-objective optimization of sieve tray hydrodynamic performance based on deep learning

Chemical Engineering Science, Journal Year: 2025, Volume and Issue: unknown, P. 121458 - 121458

Published: March 1, 2025

Language: Английский

---

A novel ensemble network based on CNN‐AM‐BiLSTM learner for temperature prediction of distillation columns

The Canadian Journal of Chemical Engineering, Journal Year: 2025, Volume and Issue: unknown

Published: March 5, 2025

Abstract In recent years, complexity has significantly increased in chemical processes where a distillation column serves as crucial unit. It is worthwhile to develop an accurate and reliable predictive model maintain the steady operation condition of column. Although data‐driven models that do not rely on any prior knowledge present promising approach, they encounter challenges associated with nonlinearity dynamic behaviour within process data. To tackle these challenges, deep learning‐based combined distilled spatiotemporal attention ensemble network (CDSAEN) proposed. The CDSAEN constructed by sequentially integrating multiple base learners, which are iteratively generated decreasing span lengths through boosting method implemented specially designed extraction evaluation function. learner, convolutional neural (CNN), mechanism (AM), bidirectional long short‐term memory (BiLSTM) utilized adaptively capture intricate features establish robust mapping relationship from inputs output. Real‐world data system plant reconstructed time series dataset subsequently fed into for training forecast temperature apparatus advance. results exhibited effectiveness reliability. Additionally, comparison six other approaches, proposed attained superior performance mean absolute error (MAE) = 0.084, root squared (RMSE) 0.108, R 2 0.974. This study can provide support maintaining stable columns processes.

Language: Английский

---

Source Tracing of Raw Material Components in Wood Vinegar Distillation Process Based on Machine Learning and Aspen Simulation

ChemEngineering, Journal Year: 2025, Volume and Issue: 9(2), P. 32 - 32

Published: March 13, 2025

As a kind of high-oxygen organic liquid produced during biomass pyrolysis, wood vinegar possesses significant industrial value due to its rich composition acetic acid, phenols, and other bioactive compounds. In this study, we explore the application advanced machine learning models in optimizing dual-column distillation process for production, such as Random Forest algorithms. Through integration Aspen Plus simulation deep learning, an adaptive control strategy is proposed enhance separation efficiency key components under varying feed conditions. The experimental results demonstrate that model exhibits superior predictive accuracy traditional decision tree methods, R2 0.9728 can be achieved phenol concentration prediction. This AI-driven system provide real-time optimization, enhancing energy efficiency, stabilizing component yields, contributing advancement sustainable practices within chemical industry. These findings are anticipated offer valuable insights into green chemistry principles with intelligent systems facilitate achievement Industry 4.0 objectives bio-based production.

Language: Английский

---

Advancements in methanol distillation system: A comprehensive overview

Process Safety and Environmental Protection, Journal Year: 2023, Volume and Issue: 199, P. 130 - 151

Published: Sept. 21, 2023

Language: Английский

---

Hyperparameter Optimization of the Machine Learning Model for Distillation Processes

International Journal of Intelligent Systems, Journal Year: 2024, Volume and Issue: 2024(1)

Published: Jan. 1, 2024

This study was conducted to enhance the efficiency of chemical process systems and address limitations conventional methods through hyperparameter optimization. Chemical processes are inherently continuous nonlinear, making stable operation challenging. The often varies significantly with operator’s level expertise, as most tasks rely on experience. To move beyond constraints traditional simulation approaches, a new machine learning‐based model developed. utilizes recurrent neural network (RNN) algorithm, which is ideal for analyzing time‐series data from systems, presenting possibilities applications in special reactions or those that complex. Hyperparameters were optimized using grid search method, optimal results confirmed when applied an actual distillation system. By proposing methodology learning optimization this research contributes solving problems previously unaddressed. Based these results, demonstrates can be effectively systems. application enables derivation unique hyperparameters tailored specificities limited control volume

Language: Английский

---

Time-series clustering approach for training data selection of a data-driven predictive model: Application to an industrial bio 2,3-butanediol distillation process

Computers & Chemical Engineering, Journal Year: 2022, Volume and Issue: 161, P. 107758 - 107758

Published: March 5, 2022

In this study, we propose a time-series clustering approach that selects optimal training data for the development of predictive models. The number clusters was set based on variation within-cluster sums squares. A model developed with selection ratio from each those clusters. Based results, regression to predict performance model. search space applied model, and were selected satisfying objective function constraints. effectiveness method is demonstrated by addressing commercial bio 2,3-butanediol distillation process. As result, reduced 49.20% compared base case without clustering. coefficient determination (R2) showed same level performance, root-mean-square error improved up 14.07%.

Language: Английский

---

A framework for environmental production of textile dyeing process using novel exhaustion-rate meter and multi-layer perceptron-based prediction model

Process Safety and Environmental Protection, Journal Year: 2023, Volume and Issue: 175, P. 99 - 110

Published: May 8, 2023

In textile industries, a lot of wastewater are discharged which one the major environmental pollution problems, because they release undesirable dye effluents. Owing to re-dyeing procedures performed meet customized color specifications, is serious problem emission large volumes wastewater. To solve problems caused by re-dyeing, right-first-time (RFT) %, rate at target quality obtained with just dyeing, must be increased considering dyeing conditions that affect product quality. Here, this study suggests framework for cleaner production process using novel exhaustion-rate meter (NERM) and multi-layer perceptron-based prediction model procedure controlling outliers. The proposed NERM measures based on absorbance solution composed measuring analysis section. metered in component through detector, performs high-resolution measurement (0.3–1.5 nm full width half maximum) via 25-μm slit 200–1100-nm wavelength range; then converted Beer's law Using NERM, an exhaustion dataset according Na2SO4 Na2CO3 consumption acquired surrogate augments data developed. MLP-based developed augmented control real-time As results, performance as regards indicated R2 values approximately 0.985 0.998, respectively, root mean squared errors (RMSE) 1.477 1.000, respectively. addition, effectiveness demonstrated application several scenarios outliers detected.

Language: Английский

---