KQKang Qin

Kang Qin

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Kang Qin has a total of 30 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Multiobjective Optimization of Integrated Green Hydrogen and Biomass Supply Networks for Vacuum Gas Oil and Bio–Oil Coprocessing in Refinery Hydrocracking Units

    by Kang Qin, Sishi Ye, Hao Wu, Mingfeng Li

    A multiobjective optimization framework is developed for the integrated green hydrogen and biomass supply network design of vacuum gas oil and bio–oil coprocessing in hydrocracking units, to support the low-carbon transformation of refinery hydrogenation processes. A mixed-integer nonlinear programming (MINLP) model is established with dual objectives of minimizing total annual cost (TAC) and total carbon emission (TCE). The ε-constraint method is adopted to generate the Pareto optimal frontier, and the evaluation function method is used to screen the optimal compromise solution balancing economic and environmental performance. Results show that the system achieves a maximum carbon emission reduction of 43.99% with only a 6.91% increase in TAC. The optimal compromise solution realizes a 39.74% TCE reduction with a negligible 0.16% TAC increase compared with the cost-optimal scenario. Green hydrogen related costs account for more than 74% of TAC, which is the dominant factor affecting the supply chain economy, while biomass transportation contributes over 60% of total emissions as the core emission reduction handle. In sensitivity analysis, the optimal supply chain configuration remains stable under fluctuations of green hydrogen price, and the effective implementation interval of carbon tax is 0–75 CNY/t CO2-eq. The proposed framework provides a quantitative basis for integrating renewable feedstocks into refinery.

  • Open Access

    Two-Stage Robust scheduling of a multi-source refinery hydrogen system under renewable uncertainty

    by Ying Chen, Han Lv, Kang Qin

    Renewable-powered water electrolysis is increasingly being considered by refineries as a means of reducing dependence on externally purchased hydrogen and lowering carbon emissions. However, fluctuations in renewable generation complicate the continuous supply of hydrogen required by refinery processes. This study proposes a two-stage optimization framework for scheduling a multi-source refinery hydrogen system under renewable-energy uncertainty. The framework combines day-ahead data-driven robust scheduling, based on adverse renewable-generation trajectories, with Intra-day rolling-horizon rescheduling. A mixed-integer linear programming model coordinates photovoltaic-powered electrolysis, dispatchable steam methane reforming, hydrogen storage, fixed internal hydrogen sources, and external gray-hydrogen procurement. A case study of a refinery with an annual processing capacity of 10 Mt shows that Intra-day rolling correction reduces external gray-hydrogen procurement by 59.68% and renewable-energy curtailment by 89.11% relative to the day-ahead robust schedule. The corresponding daily operating cost decreases by CNY 7,241. The results also show that, under the solar and wind conditions of the study site, a photovoltaic-only configuration performs better than a combined photovoltaic–wind configuration. This finding is particularly relevant to urban refineries and other sites with abundant solar resources but limited wind availability. The proposed low-carbon hydrogen configuration becomes more cost-effective than the conventional hydrogen-supply scheme when the carbon price exceeds approximately CNY 628.18/t CO2.

  • Open Access

    A Dynamic Patching LLM Framework for Adaptive Yield Prediction

    by Yang Zhang, Jinglei Pei, Kang Qin, Ting Liu, Jingbin Yang, Qinghua Wu

    Accurate yield prediction in petroleum refining is essential for optimizing industrial operations and improving economic efficiency. However, the dynamic and nonlinear nature of refining processes, together with the challenges of industrial data, complicates this task. Traditional rule-based systems lack flexibility, while deep learning methods often struggle with data noise and interpretability. We propose a dynamic patching LLM framework that adaptively constructs multi-scale temporal features, integrates process knowledge through semantic-guided prompts, and refines predictions via a self-evolving reasoning mechanism. Experiments show that our approach outperforms traditional deep learning and recent LLM-based baselines, offering a robust and interpretable solution for real-world refining applications.

  • Open Access

    ADTime: Adaptive Multivariate Time Series Forecasting Using LLMs

    by Jinglei Pei, Yang Zhang, Ting Liu, Jingbin Yang, Qinghua Wu, Kang Qin

    Large language models (LLMs) have recently demonstrated notable performance, particularly in addressing the challenge of extensive data requirements when training traditional forecasting models. However, these methods encounter significant challenges when applied to high-dimensional and domain-specific datasets. These challenges primarily arise from inability to effectively model inter-variable dependencies and capture variable-specific characteristics, leading to suboptimal performance in complex forecasting scenarios. To address these limitations, we propose ADTime, an adaptive LLM-based approach for multivariate time series forecasting. ADTime employs advanced preprocessing techniques to identify latent relationships among key variables and temporal features. Additionally, it integrates temporal alignment mechanisms and prompt-based strategies to enhance the semantic understanding of forecasting tasks by LLMs. Experimental results show that ADTime outperforms state-of-the-art methods, reducing MSE by 9.5% and MAE by 6.1% on public datasets, and by 17.1% and 13.5% on domain-specific datasets. Furthermore, zero-shot experiments on real-world refinery datasets demonstrate that ADTime exhibits stronger generalization capabilities across various transfer scenarios. These findings highlight the potential of ADTime in advancing complex, domain-specific time series forecasting tasks.


  • Open Access

    PSO-Optimized Data-Driven and Mechanism Hybrid Model to Enhance Prediction of Industrial Hydrocracking Product Yields Under Data Constraints

    by Zhenming Li, Kang Qin, Yang Zhang, Peng Yang, Yue Lou, Mingfeng Li

    The accurate prediction of hydrocracking product yields is crucial for optimizing resource allocation and improving production efficiency. However, the prediction of product flowrates in hydrocracking units often faces challenges due to insufficient data and weak correlations between input and output variables. This study proposes a hybrid framework combining a Convolutional Neural Network–Long Short-Term Memory (CNN-LSTM) model, mechanism modeling, and Particle Swarm Optimization (PSO) to address these issues. The CNN-LSTM captures spatiotemporal dependencies in operational data, while the mechanism model incorporates domain-specific physical constraints. The hybrid model is structured in both series and parallel configurations, with PSO optimizing key hyperparameters to enhance its predictive performance. The results demonstrate significant improvements in prediction accuracy, with determination coefficients (R2s) reaching 0.896 (kerosene), 0.879 (residue), 0.899 (heavy naphtha), and 0.78 (light naphtha). Shapley Additive Explanations (SHAP) and Mutual Information Coefficient (MIC) analyses highlight the mechanism model’s role in improving feature interpretability. This study underscores the efficacy of integrating kinetics modeling, deep learning, and metaheuristic optimization for complex industrial processes under data constraints, offering a robust approach to enhance hydrocracking yield prediction.


  • Open Access

    A Novel Step-by-Step Automated Heat Exchanger Network Retrofit Methodology Considering Different Heat Transfer Equipment

    by Kexin Xu, Kang Qin, Hao Wu, Robin Smith

    Improving the energy efficiency in heat exchanger networks (HENs) remains a significant industrial problem, specifically in energy-intensive operations. A particular method for such an objective is the modification of HENs at the equipment-use level, where structural changes take place and units within the network are moved, replaced and/or removed. This practice is usually known as retrofit. The objective of a retrofit is to maximize the heat recovery using the minimum modifications possible and minimum retrofit cost. Traditional retrofit techniques would normally consider one type of heat exchanger (based on the original network) with no additional design features (i.e., heat transfer enhancement technologies). The expansion of such alternatives is limited by practical use and availability of theoretical methods. In this context, the inclusion of high-performance heat exchangers such as plate heat exchangers (PHEs) has not been widely explored, even when their design and operational advantages are known. In this work, a new step-by-step automated HENs retrofit approach based on Pinch Analysis is proposed. The approach is possible to identify the best modification, its location within the network, and its cost simultaneously. Moreover, to increase energy savings, this work presents a strategy that seeks to utilize high efficiency heat exchangers such as plate heat exchangers for retrofit. A distinctive feature of this new method is the ability to handle different minimum approach temperatures, given the different types of exchangers, within the optimization of HENs. Three cases are studied using this methodology to quantify the potential benefits of including PHEs in HEN retrofits, via the analysis of the retrofit cost. Results are compared with a baseline consisting in the same network, where only Shell-and-Tube-Heat-Exchangers (STHXs) are used. In addition, the results demonstrate that this methodology is flexible enough to be applied in a wide range of retrofit problems.