Chemometrics and Intelligent Laboratory Systems最新文献_第2页

Non-invasive diagnosis of common glomerular diseases via Raman spectroscopy and machine learning: an integrated blood and urine analysis approach 通过拉曼光谱和机器学习无创诊断常见肾小球疾病：一种综合血液和尿液分析方法

IF 3.8 2区化学 Q2 AUTOMATION & CONTROL SYSTEMS

Chemometrics and Intelligent Laboratory Systems

Pub Date : 2026-01-21 DOI: 10.1016/j.chemolab.2026.105630

Mengyu Wu , Yuan Cao , Ruiyang Wang , Chongxuan Tian , Yang Li , Zunsong Wang

Background

Percutaneous renal biopsy faces three major challenges in clinical management: inherent procedural risks, inability to serially monitor disease activity, and sampling variability. These limitations underscore the demand for safer, repeatable diagnostic tools.

Objective

Our objective was to explore the potential of a liquid biopsy strategy utilizing paired blood and urine analysis via Raman spectroscopy and a 1D-CNN to facilitate the differentiation of common glomerular diseases from each other and from healthy individuals.

Methods

From January 2021 to January 2025, we collected serum and first-void morning urine from 170 biopsy-confirmed patients (81 membranous nephropathy, 36 IgA nephropathy, 33 diabetic nephropathy, 20 focal segmental glomerulosclerosis) and 21 healthy volunteers. Spectra were acquired on an Attenuated Total Reflection-8300 (ATR-8300) instrument (785 nm excitation) and preprocessed via third-order polynomial baseline correction and 13-point Savitzky–Golay smoothing. A 1D-CNN was trained on the combined spectral data; performance was assessed by accuracy, sensitivity, specificity, and Receiver Operating Characteristic - Area Under the Curve (ROC-AUC).

Results

The 1D-CNN model achieved 80.0 % accuracy, 76.2 % sensitivity, and 81.3 % specificity in five-class classification. ROC-AUCs ranged from 0.81 (FSGS) to 0.85 (IgA nephropathy), confirming robust discrimination across disease subtypes and controls. Characteristic Raman bands—e.g. phenylalanine (∼1003 cm⁻¹), Amide I (∼1655 cm⁻¹), and C–H stretching (2800–3000 cm⁻¹)—differed systematically among cohorts, reflecting underlying biochemical alterations.

Conclusions

Raman spectroscopy of paired blood and urine, coupled with deep learning, provides a rapid, label-free approach for minimally invasive classification of glomerular diseases. This integrated liquid biopsy strategy may enable early detection and precise stratification in nephrology, reducing reliance on invasive biopsy and informing personalized therapy.

背景：经皮肾活检在临床管理中面临三大挑战：固有的程序风险、无法连续监测疾病活动以及采样的可变性。这些限制强调了对更安全、可重复的诊断工具的需求。我们的目的是探索液体活检策略的潜力，利用拉曼光谱和1D-CNN对血液和尿液进行配对分析，以促进常见肾小球疾病彼此之间和健康个体之间的区分。方法从2021年1月至2025年1月，收集170例活检确诊患者（膜性肾病81例，IgA肾病36例，糖尿病肾病33例，局灶节段性肾小球硬化20例）和21名健康志愿者的血清和首次空晨尿。在衰减全反射-8300 （ATR-8300）仪器（785 nm激发）上获取光谱，并通过三阶多项式基线校正和13点Savitzky-Golay平滑进行预处理。在组合光谱数据上训练1D-CNN；通过准确性、灵敏度、特异性和受试者工作特征-曲线下面积（ROC-AUC）来评估疗效。结果1D-CNN模型在五类分类中准确率为80.0%，灵敏度为76.2%，特异度为81.3%。roc - auc范围从0.81 （FSGS）到0.85 （IgA肾病），证实了疾病亚型和对照之间的强大区别。特征拉曼波段：苯丙氨酸（~ 1003 cm−1）、酰胺I （~ 1655 cm−1）和C-H拉伸（2800-3000 cm−1）在队列中存在系统性差异，反映了潜在的生化改变。结论配对血液和尿液的拉曼光谱，结合深度学习，为肾小球疾病的微创分类提供了一种快速、无标记的方法。这种综合液体活检策略可以实现肾脏学的早期检测和精确分层，减少对侵入性活检的依赖，并为个性化治疗提供信息。

{"title":"Non-invasive diagnosis of common glomerular diseases via Raman spectroscopy and machine learning: an integrated blood and urine analysis approach","authors":"Mengyu Wu , Yuan Cao , Ruiyang Wang , Chongxuan Tian , Yang Li , Zunsong Wang","doi":"10.1016/j.chemolab.2026.105630","DOIUrl":"10.1016/j.chemolab.2026.105630","url":null,"abstract":"<div><h3>Background</h3><div>Percutaneous renal biopsy faces three major challenges in clinical management: inherent procedural risks, inability to serially monitor disease activity, and sampling variability. These limitations underscore the demand for safer, repeatable diagnostic tools.</div></div><div><h3>Objective</h3><div>Our objective was to explore the potential of a liquid biopsy strategy utilizing paired blood and urine analysis via Raman spectroscopy and a 1D-CNN to facilitate the differentiation of common glomerular diseases from each other and from healthy individuals.</div></div><div><h3>Methods</h3><div>From January 2021 to January 2025, we collected serum and first-void morning urine from 170 biopsy-confirmed patients (81 membranous nephropathy, 36 IgA nephropathy, 33 diabetic nephropathy, 20 focal segmental glomerulosclerosis) and 21 healthy volunteers. Spectra were acquired on an Attenuated Total Reflection-8300 (ATR-8300) instrument (785 nm excitation) and preprocessed via third-order polynomial baseline correction and 13-point Savitzky–Golay smoothing. A 1D-CNN was trained on the combined spectral data; performance was assessed by accuracy, sensitivity, specificity, and Receiver Operating Characteristic - Area Under the Curve (ROC-AUC).</div></div><div><h3>Results</h3><div>The 1D-CNN model achieved 80.0 % accuracy, 76.2 % sensitivity, and 81.3 % specificity in five-class classification. ROC-AUCs ranged from 0.81 (FSGS) to 0.85 (IgA nephropathy), confirming robust discrimination across disease subtypes and controls. Characteristic Raman bands—e.g. phenylalanine (∼1003 cm<sup>−1</sup>), Amide I (∼1655 cm<sup>−1</sup>), and C–H stretching (2800–3000 cm<sup>−1</sup>)—differed systematically among cohorts, reflecting underlying biochemical alterations.</div></div><div><h3>Conclusions</h3><div>Raman spectroscopy of paired blood and urine, coupled with deep learning, provides a rapid, label-free approach for minimally invasive classification of glomerular diseases. This integrated liquid biopsy strategy may enable early detection and precise stratification in nephrology, reducing reliance on invasive biopsy and informing personalized therapy.</div></div>","PeriodicalId":9774,"journal":{"name":"Chemometrics and Intelligent Laboratory Systems","volume":"270 ","pages":"Article 105630"},"PeriodicalIF":3.8,"publicationDate":"2026-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146075362","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Innovative approaches in anti-inflammatory research: QSAR-SVM-PCA, In vivo insights, and molecular docking of Laurus nobilis 抗炎研究的创新方法：QSAR-SVM-PCA，体内观察，以及月桂的分子对接

IF 3.8 2区化学 Q2 AUTOMATION & CONTROL SYSTEMS

Chemometrics and Intelligent Laboratory Systems

Pub Date : 2026-01-19 DOI: 10.1016/j.chemolab.2026.105634

Nassrine Ouafi , Asma Belkadi , Mohammed Kebir , Hichem Tahraoui , Walid Elfalleh , Fehmi Boufahja , Nasir A. Ibrahim , Nosiba S. Basher , Amin Mousavi Khaneghah , Noureddine Nesralah , Hamza Mousssa , Kahina Ighilahriz , Mustapha Mounir Bouhenna , Yacine Benguerba , Abdeltif Amrane

This study evaluates the anti-inflammatory potential of the ethanolic extract of Laurus nobilis leaves in chronic (acetic acid-induced ulcerative colitis) and acute (carrageenan-induced paw edema) inflammation models in vivo. Molecular docking studies were performed to investigate interactions between key phenolic compounds, including rutin, quercetin, and β-carotene, and inflammatory mediators such as tumor necrosis factor-alpha (TNF-α), interleukin-1 beta (IL-1β), and cyclooxygenase-2 (COX-2). The extract was obtained through cold maceration and analyzed via UPLC-MS/MS, identifying major compounds. In the chronic inflammation model, the Laurus nobilis extract significantly decreased the clinical disease index (weight loss, stool consistency, and rectal bleeding), reduced the colon weight-to-length ratio, and showed significant histological improvements compared to the control group. The acute model demonstrated a 68.84 % reduction in edema, comparable to ibuprofen's effect of 66.14 %. Molecular docking revealed strong binding affinities between β-carotene, quercetin, and rutin with COX-2, suggesting possible inhibition of inflammatory responses.

In contrast, quercetin and rutin exhibited complex interactions through hydrogen bonding with TNF-α and COX-2 active sites. A QSAR model, developed using Support Vector Machines (SVM) and Principal Component Analysis (PCA), effectively predicted the biological activity of the phenolic compounds, demonstrating high predictive capability and robustness. These findings support the traditional use of Laurus nobilis for managing inflammatory conditions and highlight its potential as an alternative therapeutic agent for inflammatory diseases, warranting further research.

本研究评估月桂叶乙醇提取物在慢性（醋酸诱导的溃疡性结肠炎）和急性（卡拉胶诱导的足跖水肿）炎症模型中的抗炎潜力。通过分子对接研究，研究了芦丁、槲皮素和β-胡萝卜素等关键酚类化合物与肿瘤坏死因子-α (TNF-α)、白细胞介素-1β (IL-1β)和环氧化酶-2 （COX-2）等炎症介质之间的相互作用。提取液经冷浸得到，通过UPLC-MS/MS进行分析，鉴定主要化合物。在慢性炎症模型中，月牙提取物显著降低了临床疾病指数（体重减轻、大便粘稠度、直肠出血），降低了结肠重长比，并与对照组相比有明显的组织学改善。急性模型显示水肿减少68.84%，与布洛芬66.14%的效果相当。分子对接发现β-胡萝卜素、槲皮素和芦丁与COX-2有很强的结合亲和力，提示可能具有抑制炎症反应的作用。相反，槲皮素和芦丁通过与TNF-α和COX-2活性位点的氢键表现出复杂的相互作用。利用支持向量机（SVM）和主成分分析（PCA）建立的QSAR模型能够有效预测酚类化合物的生物活性，具有较高的预测能力和鲁棒性。这些发现支持了月桂用于治疗炎症的传统用途，并突出了其作为炎症疾病替代治疗剂的潜力，值得进一步研究。

{"title":"Innovative approaches in anti-inflammatory research: QSAR-SVM-PCA, In vivo insights, and molecular docking of Laurus nobilis","authors":"Nassrine Ouafi , Asma Belkadi , Mohammed Kebir , Hichem Tahraoui , Walid Elfalleh , Fehmi Boufahja , Nasir A. Ibrahim , Nosiba S. Basher , Amin Mousavi Khaneghah , Noureddine Nesralah , Hamza Mousssa , Kahina Ighilahriz , Mustapha Mounir Bouhenna , Yacine Benguerba , Abdeltif Amrane","doi":"10.1016/j.chemolab.2026.105634","DOIUrl":"10.1016/j.chemolab.2026.105634","url":null,"abstract":"<div><div>This study evaluates the anti-inflammatory potential of the ethanolic extract of <em>Laurus nobilis</em> leaves in chronic (acetic acid-induced ulcerative colitis) and acute (carrageenan-induced paw edema) inflammation models <em>in vivo</em>. Molecular docking studies were performed to investigate interactions between key phenolic compounds, including rutin, quercetin, and β-carotene, and inflammatory mediators such as tumor necrosis factor-alpha (TNF-α), interleukin-1 beta (IL-1β), and cyclooxygenase-2 (COX-2). The extract was obtained through cold maceration and analyzed via UPLC-MS/MS, identifying major compounds. In the chronic inflammation model, the <em>Laurus nobilis</em> extract significantly decreased the clinical disease index (weight loss, stool consistency, and rectal bleeding), reduced the colon weight-to-length ratio, and showed significant histological improvements compared to the control group. The acute model demonstrated a 68.84 % reduction in edema, comparable to ibuprofen's effect of 66.14 %. Molecular docking revealed strong binding affinities between β-carotene, quercetin, and rutin with COX-2, suggesting possible inhibition of inflammatory responses.</div><div>In contrast, quercetin and rutin exhibited complex interactions through hydrogen bonding with TNF-α and COX-2 active sites. A QSAR model, developed using Support Vector Machines (SVM) and Principal Component Analysis (PCA), effectively predicted the biological activity of the phenolic compounds, demonstrating high predictive capability and robustness. These findings support the traditional use of <em>Laurus nobilis</em> for managing inflammatory conditions and highlight its potential as an alternative therapeutic agent for inflammatory diseases, warranting further research.</div></div>","PeriodicalId":9774,"journal":{"name":"Chemometrics and Intelligent Laboratory Systems","volume":"270 ","pages":"Article 105634"},"PeriodicalIF":3.8,"publicationDate":"2026-01-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146024965","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Integrative chemometric and intelligent modeling approaches for pollutant adsorption: Synergistic insights from experimental design, artificial intelligence (AI), and DFT applied to bisphenol A, β-naphthol, and eriochrome black 污染物吸附的综合化学计量学和智能建模方法：实验设计，人工智能（AI）和DFT应用于双酚A， β-萘酚和铬黑的协同见解

IF 3.8 2区化学 Q2 AUTOMATION & CONTROL SYSTEMS

Chemometrics and Intelligent Laboratory Systems

Pub Date : 2026-01-16 DOI: 10.1016/j.chemolab.2026.105639

Taoufiq Bouzid, Abdelali Grich, Aicha Naboulsi, Hicham Yazid, Ali Elbasraoui, Abdelmajid Regti, Mamoune El Himri, Mohammadine El Haddad

Water pollution has emerged as one of the most pressing environmental challenges in recent years. Various solutions have been investigated to mitigate this issue, and among them, adsorption has proven to be an innovative and efficient technique for removing pollutants from water. In this review, we highlight the combined use of Artificial Intelligence (AI), experimental design, and Density Functional Theory (DFT) calculations three powerful tools whose integration in adsorption studies has not yet been reported in the literature.

Within the scope of AI, we explore the selection and evaluation of different models applied to adsorption processes. Specifically, we discuss Artificial Neural Networks (ANN), k-Nearest Neighbors (KNN), Random Forests (RF), and Linear Regression. Overall, ANN emerged as the most effective approach, as it can account for multiple factors influencing adsorption simultaneously. In this section, we also define AI concepts, describe the models, explain data collection strategies, and outline validation methods to ensure accurate prediction of adsorption performance.

Regarding experimental design, we compare two approaches: the full factorial design and the central composite design. We demonstrate the advantages of the central composite design in optimizing adsorption conditions. This part also details how experimental plans are structured, validated, and how outputs such as 3D surface plots and other response analyses can be leveraged to extract valuable insights.

Finally, we examine the role of DFT calculations, emphasizing their ability to provide a deeper understanding of adsorption mechanisms at the molecular level.

To illustrate the practical application of these integrated methodologies, we present case studies on three representative pollutants: Bisphenol A, β-naphthol, and Eriochrome Black T, which exemplify the diversity of contaminants that can be addressed through this approach.

近年来，水污染已成为最紧迫的环境挑战之一。人们已经研究了各种解决方案来缓解这一问题，其中，吸附已被证明是一种从水中去除污染物的创新和有效技术。在这篇综述中，我们强调了人工智能（AI）、实验设计和密度泛函理论（DFT）计算这三种强大工具在吸附研究中的结合使用，这些工具在文献中尚未报道。在人工智能的范围内，我们探索了应用于吸附过程的不同模型的选择和评估。具体来说，我们讨论了人工神经网络（ANN）、k近邻（KNN）、随机森林（RF）和线性回归。总的来说，人工神经网络是最有效的方法，因为它可以同时考虑影响吸附的多种因素。在本节中，我们还定义了AI概念，描述了模型，解释了数据收集策略，并概述了验证方法，以确保准确预测吸附性能。在实验设计方面，我们比较了两种方法：全因子设计和中心复合设计。我们证明了中心复合设计在优化吸附条件方面的优势。本部分还详细介绍了实验计划的结构，验证以及如何利用诸如3D表面图和其他响应分析等输出来提取有价值的见解。最后，我们研究了DFT计算的作用，强调了它们在分子水平上提供对吸附机制更深层次理解的能力。为了说明这些综合方法的实际应用，我们介绍了三种代表性污染物的案例研究：双酚A、β-萘酚和铬黑T，这说明了通过这种方法可以解决的污染物的多样性。

{"title":"Integrative chemometric and intelligent modeling approaches for pollutant adsorption: Synergistic insights from experimental design, artificial intelligence (AI), and DFT applied to bisphenol A, β-naphthol, and eriochrome black","authors":"Taoufiq Bouzid, Abdelali Grich, Aicha Naboulsi, Hicham Yazid, Ali Elbasraoui, Abdelmajid Regti, Mamoune El Himri, Mohammadine El Haddad","doi":"10.1016/j.chemolab.2026.105639","DOIUrl":"10.1016/j.chemolab.2026.105639","url":null,"abstract":"<div><div>Water pollution has emerged as one of the most pressing environmental challenges in recent years. Various solutions have been investigated to mitigate this issue, and among them, adsorption has proven to be an innovative and efficient technique for removing pollutants from water. In this review, we highlight the combined use of Artificial Intelligence (AI), experimental design, and Density Functional Theory (DFT) calculations three powerful tools whose integration in adsorption studies has not yet been reported in the literature.</div><div>Within the scope of AI, we explore the selection and evaluation of different models applied to adsorption processes. Specifically, we discuss Artificial Neural Networks (ANN), k-Nearest Neighbors (KNN), Random Forests (RF), and Linear Regression. Overall, ANN emerged as the most effective approach, as it can account for multiple factors influencing adsorption simultaneously. In this section, we also define AI concepts, describe the models, explain data collection strategies, and outline validation methods to ensure accurate prediction of adsorption performance.</div><div>Regarding experimental design, we compare two approaches: the full factorial design and the central composite design. We demonstrate the advantages of the central composite design in optimizing adsorption conditions. This part also details how experimental plans are structured, validated, and how outputs such as 3D surface plots and other response analyses can be leveraged to extract valuable insights.</div><div>Finally, we examine the role of DFT calculations, emphasizing their ability to provide a deeper understanding of adsorption mechanisms at the molecular level.</div><div>To illustrate the practical application of these integrated methodologies, we present case studies on three representative pollutants: Bisphenol A, β-naphthol, and Eriochrome Black T, which exemplify the diversity of contaminants that can be addressed through this approach.</div></div>","PeriodicalId":9774,"journal":{"name":"Chemometrics and Intelligent Laboratory Systems","volume":"270 ","pages":"Article 105639"},"PeriodicalIF":3.8,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146025044","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Eyes on every node: Adaptive neighborhood perception for spatiotemporal data intelligent modeling and its industrial application 着眼每一个节点：时空数据智能建模的自适应邻域感知及其工业应用

IF 3.8 2区化学 Q2 AUTOMATION & CONTROL SYSTEMS

Chemometrics and Intelligent Laboratory Systems

Pub Date : 2026-01-13 DOI: 10.1016/j.chemolab.2026.105633

Yalin Wang , Ruikai Yang , Chenliang Liu , Zhongmei Li , Yijing Fang , Weihua Gui

Accurate online prediction of key quality variables is a guiding indicator for process optimization and stable operation in industrial processes. Due to the continuous and occasionally abrupt nature of industrial processes, industrial data often exhibit complex spatiotemporal coupling characteristics across long-range spatial and adjacent temporal dimensions. In particular, the dynamic variation of local spatiotemporal neighborhood space makes it challenging for traditional methods to capture these patterns. To address this issue, this paper proposes a novel neighborhood attention-aware spatiotemporal manifold autoencoder (NA-STMAE) model for soft sensor modeling of quality variables, which is designed to learn adaptive correlations within spatial and temporal neighborhoods of industrial data. Specifically, a novel attention-based neighborhood computing mode is designed to dynamically allocate weights among local samples, enabling adaptive perception and refinement of neighborhood relationships. Based on this, an attention-aware spatiotemporal neighborhood feature extraction module is developed to learn local spatiotemporal dependencies, thereby enhancing the predictive performance of the proposed soft sensor model. Finally, extensive experiments were conducted on two industrial processes to validate the effectiveness of the proposed model. Experimental results demonstrate that the proposed model outperforms several mainstream soft sensor models in prediction tasks. Moreover, ablation experiments further confirm the critical role of dynamic weight allocation in capturing both temporal and spatial dimensions.

关键质量变量的准确在线预测是工业过程优化和稳定运行的指导性指标。由于工业过程的连续性和偶尔的突发性，工业数据往往表现出复杂的时空耦合特征，跨越远距离空间和相邻时间维度。特别是局部时空邻域空间的动态变化使得传统方法难以捕捉这些模式。为了解决这一问题，本文提出了一种新的邻域注意力感知时空流形自编码器（NA-STMAE）模型，用于质量变量的软传感器建模，该模型旨在学习工业数据的时空邻域内的自适应相关性。具体而言，设计了一种新的基于注意力的邻域计算模式，在局部样本之间动态分配权重，实现邻域关系的自适应感知和细化。在此基础上，开发了注意感知时空邻域特征提取模块，学习局部时空依赖关系，从而提高了软传感器模型的预测性能。最后，在两个工业过程中进行了大量的实验来验证所提出模型的有效性。实验结果表明，该模型在预测任务方面优于几种主流软测量模型。此外，消融实验进一步证实了动态权重分配在捕获时间和空间维度方面的关键作用。

{"title":"Eyes on every node: Adaptive neighborhood perception for spatiotemporal data intelligent modeling and its industrial application","authors":"Yalin Wang , Ruikai Yang , Chenliang Liu , Zhongmei Li , Yijing Fang , Weihua Gui","doi":"10.1016/j.chemolab.2026.105633","DOIUrl":"10.1016/j.chemolab.2026.105633","url":null,"abstract":"<div><div>Accurate online prediction of key quality variables is a guiding indicator for process optimization and stable operation in industrial processes. Due to the continuous and occasionally abrupt nature of industrial processes, industrial data often exhibit complex spatiotemporal coupling characteristics across long-range spatial and adjacent temporal dimensions. In particular, the dynamic variation of local spatiotemporal neighborhood space makes it challenging for traditional methods to capture these patterns. To address this issue, this paper proposes a novel neighborhood attention-aware spatiotemporal manifold autoencoder (NA-STMAE) model for soft sensor modeling of quality variables, which is designed to learn adaptive correlations within spatial and temporal neighborhoods of industrial data. Specifically, a novel attention-based neighborhood computing mode is designed to dynamically allocate weights among local samples, enabling adaptive perception and refinement of neighborhood relationships. Based on this, an attention-aware spatiotemporal neighborhood feature extraction module is developed to learn local spatiotemporal dependencies, thereby enhancing the predictive performance of the proposed soft sensor model. Finally, extensive experiments were conducted on two industrial processes to validate the effectiveness of the proposed model. Experimental results demonstrate that the proposed model outperforms several mainstream soft sensor models in prediction tasks. Moreover, ablation experiments further confirm the critical role of dynamic weight allocation in capturing both temporal and spatial dimensions.</div></div>","PeriodicalId":9774,"journal":{"name":"Chemometrics and Intelligent Laboratory Systems","volume":"270 ","pages":"Article 105633"},"PeriodicalIF":3.8,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145976380","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Robust integrated chemometric driven approach for the analysis of cilnidipine and chlorthalidone in biological and pharmaceutical matrices 生物和药物基质中西尼地平和氯噻酮的鲁棒综合化学计量驱动分析方法

IF 3.8 2区化学 Q2 AUTOMATION & CONTROL SYSTEMS

Chemometrics and Intelligent Laboratory Systems

Pub Date : 2026-01-13 DOI: 10.1016/j.chemolab.2026.105637

Suraj R. Chaudhari , Atul A. Shirkhedkar

Cilnidipine (CIL) and chlorthalidone (CHL), both antihypertensive agents are approved for combined regimens for the management of hypertension. A thorough evaluation of their intrinsic stability and content levels in commercially available preparations and biological samples requires a simple and reliable analytical approach. Herein, computational approaches, stability investigations, and retrospective analysis of content uniformity results were explored to support the robustness and long-term suitability of the proposed protocol for routine application. Therefore, this experiment established an ultra-fluid liquid chromatography with diode array detection (UFLC-PDA) for simultaneous separation and quantification of CIL and CHL in Cilacar C, Nexovas CH tablets, and biological matrices. The applicability of the established protocol was confirmed with the ICH Q2 (R2), Q1A (R2), and Q1B recommendations. Analytes were extracted in a simple single step and analyzed using a rapid resolution ZORBAX Eclipse C₁₈ column (4.6 mm internal diameter

\times

100 mm length with 3.5 μm particle size), maintained at 33

°C

as column oven temperature. The resolution was observed using binary gradient elution at 0.5 mL/min with a solvent system comprising H₂O: ACN (25.85:74.15 % v/v). CHL and CIL were detected at a retention time (t_R) of 2.221 ± 0.003 min and 4.435 ± 0.011 min, with a total run time <8.0 min. The proposed protocol demonstrates outstanding specificity and sensitivity, offering a systematic platform for developing and refining knowledge related to a UFLC-PDA procedure. Moreover, it shows a comprehensive understanding of the procedure to meet the requirements specified in ICH Q14.

西尼地平（CIL）和氯噻酮（CHL）这两种抗高血压药物被批准用于高血压治疗的联合治疗方案。全面评估它们在市售制剂和生物样品中的内在稳定性和含量水平需要一种简单可靠的分析方法。本文探讨了计算方法、稳定性调查和内容均匀性结果的回顾性分析，以支持拟议方案在常规应用中的鲁棒性和长期适用性。因此，本实验建立了二极管阵列检测超流体液相色谱（UFLC-PDA）同时分离定量Cilacar C、Nexovas CH片剂和生物基质中CIL和CHL的方法。既定方案的适用性根据ICH Q2 （R2）、Q1A （R2）和Q1B建议得到确认。色谱柱为ZORBAX Eclipse C18柱（4.6 mm内径× 100 mm长，3.5 μm粒度），柱箱温度为33℃。以H2O: ACN （25.85: 74.15% v/v）为溶剂体系，以0.5 mL/min的速度进行二元梯度洗脱。CHL和CIL的滞留时间（tR）分别为2.221±0.003 min和4.435±0.011 min，总运行时间为8.0 min。该方案具有突出的特异性和敏感性，为开发和完善与UFLC-PDA程序相关的知识提供了系统的平台。此外，它显示了对程序的全面理解，以满足ICH Q14规定的要求。

{"title":"Robust integrated chemometric driven approach for the analysis of cilnidipine and chlorthalidone in biological and pharmaceutical matrices","authors":"Suraj R. Chaudhari , Atul A. Shirkhedkar","doi":"10.1016/j.chemolab.2026.105637","DOIUrl":"10.1016/j.chemolab.2026.105637","url":null,"abstract":"<div><div>Cilnidipine (CIL) and chlorthalidone (CHL), both antihypertensive agents are approved for combined regimens for the management of hypertension. A thorough evaluation of their intrinsic stability and content levels in commercially available preparations and biological samples requires a simple and reliable analytical approach. Herein, computational approaches, stability investigations, and retrospective analysis of content uniformity results were explored to support the robustness and long-term suitability of the proposed protocol for routine application. Therefore, this experiment established an ultra-fluid liquid chromatography with diode array detection (UFLC-PDA) for simultaneous separation and quantification of CIL and CHL in Cilacar C, Nexovas CH tablets, and biological matrices. The applicability of the established protocol was confirmed with the ICH Q2 (R2), Q1A (R2), and Q1B recommendations. Analytes were extracted in a simple single step and analyzed using a rapid resolution ZORBAX Eclipse C<sub>18</sub> column (4.6 mm internal diameter <span><math><mrow><mo>×</mo></mrow></math></span> 100 mm length with 3.5 μm particle size), maintained at 33 <span><math><mrow><mo>°C</mo></mrow></math></span> as column oven temperature. The resolution was observed using binary gradient elution at 0.5 mL/min with a solvent system comprising H<sub>2</sub>O: ACN (25.85:74.15 % <em>v/v</em>). CHL and CIL were detected at a retention time (t<sub>R</sub>) of 2.221 ± 0.003 min and 4.435 ± 0.011 min, with a total run time <8.0 min. The proposed protocol demonstrates outstanding specificity and sensitivity, offering a systematic platform for developing and refining knowledge related to a UFLC-PDA procedure. Moreover, it shows a comprehensive understanding of the procedure to meet the requirements specified in ICH Q14.</div></div>","PeriodicalId":9774,"journal":{"name":"Chemometrics and Intelligent Laboratory Systems","volume":"270 ","pages":"Article 105637"},"PeriodicalIF":3.8,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146024966","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A mechanistic causality-guided robust dynamical probabilistic latent variable regression model and its application to soft sensing of continuous chemical processes 一种机械因果导向的稳健动态概率潜变量回归模型及其在连续化工过程软测量中的应用

IF 3.8 2区化学 Q2 AUTOMATION & CONTROL SYSTEMS

Chemometrics and Intelligent Laboratory Systems

Pub Date : 2026-01-13 DOI: 10.1016/j.chemolab.2026.105638

Wenxue Han , Ziteng Zuo , Xiangjing Zhang , Lan Zhang , Weiming Shao

Precisely predicting quality variables is crucial for advanced process control and real-time optimization in continuous chemical processes. Soft sensing technology is utilized for this task due to its advantages of real-time capability and low cost. Dynamical probabilistic latent variable regression (DPLVR) models for soft sensing modeling have attracted increasing attention, owing to their superior feature extraction capability. Nevertheless, the DPLVR-based soft sensing methods only account for variable correlations while neglecting the underlying causal mechanisms. Current research on causal methods primarily focuses on the selection of causal variables and the construction of causal graphs, failing to effectively integrate the causal priors that reflect the underlying mechanisms of chemical processes. In addition, outliers in chemical data further degrade the prediction accuracy of soft sensors, making them inadequate for practical production requirements. Given the above problems, a novel mechanistic causality-guided robust DPLVR (MCR-DPLVR) model is proposed for predicting the quality variables. In the MCR-DPLVR, the mechanistic causality knowledge is used to identify the causal mechanisms among different types of variables, and the Student’s

t

distribution is utilized to enhance the model’s robustness against outliers. Subsequently, an efficient semi-supervised training algorithm is developed to train the MCR-DPLVR based on the expectation–maximization algorithm. Furthermore, the effectiveness of the MCR-DPLVR is verified by a synthetic numerical case and an actual hydrogen production process, which exhibits the superiority of the MCR-DPLVR in comparison to several cutting-edge methods.

在连续化工过程中，精确预测质量变量对先进的过程控制和实时优化至关重要。软测量技术具有实时性好、成本低的优点，可用于该任务。动态概率潜变量回归（DPLVR）模型由于其优越的特征提取能力而越来越受到人们的关注。然而，基于dplvr的软测量方法只考虑变量相关性，而忽略了潜在的因果机制。目前对因果方法的研究主要集中在因果变量的选择和因果图的构建上，未能有效整合反映化学过程潜在机制的因果先验。此外，化学数据中的异常值进一步降低了软传感器的预测精度，使其无法满足实际生产要求。针对上述问题，提出了一种新的机制因果导向鲁棒DPLVR （MCR-DPLVR）模型来预测质量变量。在MCR-DPLVR中，利用机制因果关系知识来识别不同类型变量之间的因果机制，并利用Student 's t分布来增强模型对异常值的稳健性。随后，提出了一种基于期望最大化算法的高效半监督训练算法来训练MCR-DPLVR。最后，通过综合数值算例和实际制氢过程验证了MCR-DPLVR的有效性，表明了MCR-DPLVR与几种前沿制氢方法相比的优越性。

{"title":"A mechanistic causality-guided robust dynamical probabilistic latent variable regression model and its application to soft sensing of continuous chemical processes","authors":"Wenxue Han , Ziteng Zuo , Xiangjing Zhang , Lan Zhang , Weiming Shao","doi":"10.1016/j.chemolab.2026.105638","DOIUrl":"10.1016/j.chemolab.2026.105638","url":null,"abstract":"<div><div>Precisely predicting quality variables is crucial for advanced process control and real-time optimization in continuous chemical processes. Soft sensing technology is utilized for this task due to its advantages of real-time capability and low cost. Dynamical probabilistic latent variable regression (DPLVR) models for soft sensing modeling have attracted increasing attention, owing to their superior feature extraction capability. Nevertheless, the DPLVR-based soft sensing methods only account for variable correlations while neglecting the underlying causal mechanisms. Current research on causal methods primarily focuses on the selection of causal variables and the construction of causal graphs, failing to effectively integrate the causal priors that reflect the underlying mechanisms of chemical processes. In addition, outliers in chemical data further degrade the prediction accuracy of soft sensors, making them inadequate for practical production requirements. Given the above problems, a novel mechanistic causality-guided robust DPLVR (MCR-DPLVR) model is proposed for predicting the quality variables. In the MCR-DPLVR, the mechanistic causality knowledge is used to identify the causal mechanisms among different types of variables, and the Student’s <span><math><mi>t</mi></math></span> distribution is utilized to enhance the model’s robustness against outliers. Subsequently, an efficient semi-supervised training algorithm is developed to train the MCR-DPLVR based on the expectation–maximization algorithm. Furthermore, the effectiveness of the MCR-DPLVR is verified by a synthetic numerical case and an actual hydrogen production process, which exhibits the superiority of the MCR-DPLVR in comparison to several cutting-edge methods.</div></div>","PeriodicalId":9774,"journal":{"name":"Chemometrics and Intelligent Laboratory Systems","volume":"270 ","pages":"Article 105638"},"PeriodicalIF":3.8,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145976381","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Transformer-based sequence-to-sequence soft sensor using missing data in industrial processes 基于变压器的工业过程中缺失数据的序列对序列软传感器

IF 3.8 2区化学 Q2 AUTOMATION & CONTROL SYSTEMS

Chemometrics and Intelligent Laboratory Systems

Pub Date : 2026-01-09 DOI: 10.1016/j.chemolab.2026.105631

Yi Liu , Jiefei Gao , Jianwen Shao , Jian Wu , Mingwei Jia , Qiao Liu

Handling missing process data is a basic requirement for soft sensor development in process systems engineering. In strongly dynamic and nonlinear operations, imputers that ignore information carried by key performance indicators often yield biased completions and deteriorate soft sensing performance. To address this issue, a transformer-based sequence-to-sequence model (Trans-S2S) is proposed that performs imputation and prediction jointly within an end-to-end framework. The self-attention mechanism can directly model long-term dependencies at different times and is less sensitive to missing values. A long short-term memory then imputes missing values while capturing short-term nonlinearity. For the prediction, a long short-term memory is used to predict key performance indicators from the completed data. Both modules are trained jointly, allowing the soft sensing task to guide imputation. Across missing rate scenarios, Trans-S2S reduces soft sensing errors by 22.41 %–63.75 % on the penicillin fermentation benchmark and by 4.76 %–30.00 % on the cement production dataset, compared with baseline methods. Explanation analysis on the penicillin case further indicates that the imputations remain context-consistent and agree with known process mechanisms.

处理缺失过程数据是过程系统工程中软传感器开发的基本要求。在强动态和非线性操作中，忽略关键性能指标所携带信息的输入器通常会产生偏差完井并降低软测量性能。为了解决这个问题，提出了一种基于变压器的序列到序列模型（Trans-S2S），该模型在端到端框架内联合执行imputation和预测。自注意机制可以直接对不同时间的长期依赖关系进行建模，对缺失值的敏感性较低。长短期记忆然后在捕捉短期非线性的同时将缺失的值归因。对于预测，使用长短期记忆从完成的数据中预测关键性能指标。这两个模块是联合训练的，允许软测量任务指导插补。在缺失率情况下，与基线方法相比，Trans-S2S在青霉素发酵基准上减少了22.41% - 63.75%的软测量误差，在水泥生产数据集上减少了4.76% - 30.00%的软测量误差。对青霉素病例的解释分析进一步表明，归因保持上下文一致，并与已知的过程机制一致。

{"title":"Transformer-based sequence-to-sequence soft sensor using missing data in industrial processes","authors":"Yi Liu , Jiefei Gao , Jianwen Shao , Jian Wu , Mingwei Jia , Qiao Liu","doi":"10.1016/j.chemolab.2026.105631","DOIUrl":"10.1016/j.chemolab.2026.105631","url":null,"abstract":"<div><div>Handling missing process data is a basic requirement for soft sensor development in process systems engineering. In strongly dynamic and nonlinear operations, imputers that ignore information carried by key performance indicators often yield biased completions and deteriorate soft sensing performance. To address this issue, a transformer-based sequence-to-sequence model (<em>Trans</em>-S2S) is proposed that performs imputation and prediction jointly within an end-to-end framework. The self-attention mechanism can directly model long-term dependencies at different times and is less sensitive to missing values. A long short-term memory then imputes missing values while capturing short-term nonlinearity. For the prediction, a long short-term memory is used to predict key performance indicators from the completed data. Both modules are trained jointly, allowing the soft sensing task to guide imputation. Across missing rate scenarios, <em>Trans</em>-S2S reduces soft sensing errors by 22.41 %–63.75 % on the penicillin fermentation benchmark and by 4.76 %–30.00 % on the cement production dataset, compared with baseline methods. Explanation analysis on the penicillin case further indicates that the imputations remain context-consistent and agree with known process mechanisms.</div></div>","PeriodicalId":9774,"journal":{"name":"Chemometrics and Intelligent Laboratory Systems","volume":"270 ","pages":"Article 105631"},"PeriodicalIF":3.8,"publicationDate":"2026-01-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146074876","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

An enhanced stacking ensemble learning strategy for product quality estimation in complex industrial processes considering multi-timescale data 考虑多时间尺度数据的复杂工业过程产品质量估计的增强叠加集成学习策略

IF 3.8 2区化学 Q2 AUTOMATION & CONTROL SYSTEMS

Chemometrics and Intelligent Laboratory Systems

Pub Date : 2026-01-09 DOI: 10.1016/j.chemolab.2026.105635

Weihang Sun , Xin Jin , Sen Xie , Rui Wang

The cascaded structure involving multi-process and multi-equipment in complex industrial processes leads to time delays for product quality detection, impacting the optimization and control accuracy of industrial production. To tackle this challenge, an enhanced stacking ensemble learning model for quality estimation in complex industrial processes is developed in this paper. First, features from high-dimensional data are extracted through convolutional neural network (CNN). Subsequently, the base learner is constructed, and the meta-learning model is determined by the optimal weight coefficient selection strategy. In addition, the quality estimation error of the meta-learner is calculated and the error estimation is performed by random forest. Following this, the error compensation is applied to the meta-learner for enhancing the estimation accuracy. Through an actual industrial evaporation case for alumina production, it is demonstrated that, compared with other state-of-the-art estimation models, the estimation model present in this paper not only reduces error level but also significantly improves the estimation reliability, thereby fulfilling the operational requirements of the process industry.

复杂工业过程中涉及多工序、多设备的级联结构导致了产品质量检测的时间延迟，影响了工业生产的优化和控制精度。为了解决这一问题，本文提出了一种用于复杂工业过程质量估计的增强型叠加集成学习模型。首先，通过卷积神经网络（CNN）对高维数据进行特征提取。然后，构建基础学习器，通过最优权系数选择策略确定元学习模型。此外，计算了元学习器的质量估计误差，并用随机森林进行误差估计。在此基础上，对元学习器进行误差补偿，提高估计精度。通过一个氧化铝生产的实际工业蒸发案例表明，与其他最先进的估算模型相比，本文提出的估算模型不仅降低了误差水平，而且显著提高了估算可靠性，满足了过程工业的运行要求。

{"title":"An enhanced stacking ensemble learning strategy for product quality estimation in complex industrial processes considering multi-timescale data","authors":"Weihang Sun , Xin Jin , Sen Xie , Rui Wang","doi":"10.1016/j.chemolab.2026.105635","DOIUrl":"10.1016/j.chemolab.2026.105635","url":null,"abstract":"<div><div>The cascaded structure involving multi-process and multi-equipment in complex industrial processes leads to time delays for product quality detection, impacting the optimization and control accuracy of industrial production. To tackle this challenge, an enhanced stacking ensemble learning model for quality estimation in complex industrial processes is developed in this paper. First, features from high-dimensional data are extracted through convolutional neural network (CNN). Subsequently, the base learner is constructed, and the meta-learning model is determined by the optimal weight coefficient selection strategy. In addition, the quality estimation error of the meta-learner is calculated and the error estimation is performed by random forest. Following this, the error compensation is applied to the meta-learner for enhancing the estimation accuracy. Through an actual industrial evaporation case for alumina production, it is demonstrated that, compared with other state-of-the-art estimation models, the estimation model present in this paper not only reduces error level but also significantly improves the estimation reliability, thereby fulfilling the operational requirements of the process industry.</div></div>","PeriodicalId":9774,"journal":{"name":"Chemometrics and Intelligent Laboratory Systems","volume":"269 ","pages":"Article 105635"},"PeriodicalIF":3.8,"publicationDate":"2026-01-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145972956","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A Bayesian network-based robust framework for determining density, viscosity, and thermal conductivity of near-critical-state CO2 applied in CCUS 一个基于贝叶斯网络的鲁棒框架，用于确定CCUS中近临界状态CO2的密度、粘度和导热性

IF 3.8 2区化学 Q2 AUTOMATION & CONTROL SYSTEMS

Chemometrics and Intelligent Laboratory Systems

Pub Date : 2026-01-09 DOI: 10.1016/j.chemolab.2026.105636

Bingtao Zhao , Lu Ding , Yaxin Su

Accurate modeling of the thermophysical properties of CO₂ in the region around the critical point (RACP) or at the near-critical state is crucial for performance assessment and process design in CO₂ capture, utilization, and storage. Despite its importance, significant challenges persist because of sharp fluctuations in these properties induced by critical effects, which directly influence flow behavior, interfacial tension, and process performance. To address this issue, we develop a Bayesian regularized neural network (BRNN)-based robust framework to predict the density, viscosity, and thermal conductivity of CO₂ within the RACP. Initial steps involve constructing a backpropagation neural network facilitated by the Kennard-Stone algorithm for data partitioning. A comprehensive analysis is conducted to evaluate the impacts of training algorithms, the number of neurons in the hidden layer, and optimization methods on network performance. By refining the training procedure and optimizing weights and thresholds using the genetic algorithm, we ultimately propose a more accurate model named GA-BRNN. This model demonstrates superior generalization capabilities when compared to traditional correlations and other machine learning models for the prediction of CO₂ properties in the RACP, yielding the mean squared error of 2.0484 × 10⁻⁴ (R² = 0.9635) for density, 1.8680 (R² = 0.9743) for viscosity, and 5.2196 (R² = 0.9900) for thermal conductivity. The findings may provide a positive reference for modeling the thermophysical properties of near-critical-state CO₂ applied in the processes related to carbon capture, utilization, and storage.

准确模拟二氧化碳在临界点（RACP）附近或近临界状态下的热物理性质，对二氧化碳捕集、利用和封存过程的性能评估和工艺设计至关重要。尽管它很重要，但由于临界效应导致这些特性急剧波动，直接影响流动行为、界面张力和工艺性能，因此仍然存在重大挑战。为了解决这个问题，我们开发了一个基于贝叶斯正则化神经网络（BRNN）的鲁棒框架来预测RACP内CO2的密度、粘度和导热性。最初的步骤包括构建一个反向传播神经网络，由Kennard-Stone算法促进数据分区。综合分析了训练算法、隐层神经元数量、优化方法对网络性能的影响。通过改进训练过程并使用遗传算法优化权值和阈值，我们最终提出了一个更精确的模型，命名为GA-BRNN。与传统的相关性和其他机器学习模型相比，该模型在预测RACP中CO2性质方面表现出了优越的泛化能力，密度的均方误差为2.0484 × 10−4 (R2 = 0.9635)，粘度的均方误差为1.8680 (R2 = 0.9743)，导热系数的均方误差为5.2196 （R2 = 0.9900）。研究结果可为碳捕获、利用和封存过程中近临界态CO2的热物理性质建模提供积极参考。

{"title":"A Bayesian network-based robust framework for determining density, viscosity, and thermal conductivity of near-critical-state CO2 applied in CCUS","authors":"Bingtao Zhao , Lu Ding , Yaxin Su","doi":"10.1016/j.chemolab.2026.105636","DOIUrl":"10.1016/j.chemolab.2026.105636","url":null,"abstract":"<div><div>Accurate modeling of the thermophysical properties of CO<sub>2</sub> in the region around the critical point (RACP) or at the near-critical state is crucial for performance assessment and process design in CO<sub>2</sub> capture, utilization, and storage. Despite its importance, significant challenges persist because of sharp fluctuations in these properties induced by critical effects, which directly influence flow behavior, interfacial tension, and process performance. To address this issue, we develop a Bayesian regularized neural network (BRNN)-based robust framework to predict the density, viscosity, and thermal conductivity of CO<sub>2</sub> within the RACP. Initial steps involve constructing a backpropagation neural network facilitated by the Kennard-Stone algorithm for data partitioning. A comprehensive analysis is conducted to evaluate the impacts of training algorithms, the number of neurons in the hidden layer, and optimization methods on network performance. By refining the training procedure and optimizing weights and thresholds using the genetic algorithm, we ultimately propose a more accurate model named GA-BRNN. This model demonstrates superior generalization capabilities when compared to traditional correlations and other machine learning models for the prediction of CO<sub>2</sub> properties in the RACP, yielding the mean squared error of 2.0484 × 10<sup>−4</sup> (R<sup>2</sup> = 0.9635) for density, 1.8680 (R<sup>2</sup> = 0.9743) for viscosity, and 5.2196 (R<sup>2</sup> = 0.9900) for thermal conductivity. The findings may provide a positive reference for modeling the thermophysical properties of near-critical-state CO<sub>2</sub> applied in the processes related to carbon capture, utilization, and storage.</div></div>","PeriodicalId":9774,"journal":{"name":"Chemometrics and Intelligent Laboratory Systems","volume":"269 ","pages":"Article 105636"},"PeriodicalIF":3.8,"publicationDate":"2026-01-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145972955","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Multimodal fusion of CT features and density for rapid prediction of raw-coal ash CT特征与密度的多模态融合用于原煤灰分快速预测

IF 3.8 2区化学 Q2 AUTOMATION & CONTROL SYSTEMS

Chemometrics and Intelligent Laboratory Systems

Pub Date : 2026-01-06 DOI: 10.1016/j.chemolab.2026.105628

Shuxian Su , Peixian Geng , Jiashan Yang , Gansu Zhang , Gehang Xue , Liang Dong , Zhaolin Lu

Ash content is a key quality index in coal preparation. To overcome the time-consuming, labor-intensive nature of offline assays, the interference sensitivity of online measurements, and the surface-only information of vision/spectroscopy, this study proposes a multimodal deep-learning framework for fast and accurate raw-coal ash prediction. First, a CT-based data acquisition system was designed to synchronously collect CT image slices of raw-coal particles and particle-density information. Through quantitative analysis of the association between density and ash content, the decisive role of density – as a key indicator of coal properties – in ash prediction is revealed. By further mining the frequency-domain and spatial features of coal CT images, and based on these findings, a multimodal approach is formulated that fuses CT features with density. The model adopts a three-branch architecture: an improved EfficientNet-B0 branch learns spatial cues, an AshFormer branch captures frequency patterns related to mineral distribution and microstructural discontinuities, and a multilayer perceptron encodes density. Cross-modal attention achieves deep fusion and complementarity across modalities, and a KAN-based regression head outputs ash content. On industrial data, the proposed method attains MAPE = 0.0468, RMSE = 0.0573, and

R^{2} = 97.0 %

, outperforming single-modality image models (

Δ M A P E = - 0.0096

,

Δ R M S E = - 0.0413

,

Δ R^{2} = + 4.48

percentage points). These results demonstrate the advantage of multimodal fusion in improving the accuracy and generalization of coal-quality analysis and provide a new approach for rapid analysis under small-sample conditions.

灰分是选煤过程中重要的质量指标。为了克服离线分析的耗时、劳动密集型、在线测量的干扰敏感性以及视觉/光谱的表面信息，本研究提出了一种多模态深度学习框架，用于快速、准确地预测原煤灰分。首先，设计了基于CT的数据采集系统，同步采集原煤颗粒CT图像切片和颗粒密度信息；通过定量分析密度与灰分之间的关系，揭示了密度作为煤质的关键指标在灰分预测中的决定性作用。通过进一步挖掘煤炭CT图像的频域和空间特征，并基于这些发现，制定了一种融合CT特征和密度的多模态方法。该模型采用三分支架构：改进的EfficientNet-B0分支学习空间线索，AshFormer分支捕获与矿物分布和微观结构不连续相关的频率模式，多层感知器编码密度。跨模态注意实现了模态间的深度融合和互补，基于kan的回归头输出灰分含量。在工业数据上，本文方法的MAPE= 0.0468， RMSE= 0.0573, R2=97.0%，优于单模态图像模型（ΔMAPE=−0.0096，ΔRMSE=−0.0413，ΔR2=+4.48个百分点）。这些结果表明了多模态融合在提高煤质分析精度和泛化方面的优势，为小样本条件下的快速分析提供了一种新的方法。

{"title":"Multimodal fusion of CT features and density for rapid prediction of raw-coal ash","authors":"Shuxian Su , Peixian Geng , Jiashan Yang , Gansu Zhang , Gehang Xue , Liang Dong , Zhaolin Lu","doi":"10.1016/j.chemolab.2026.105628","DOIUrl":"10.1016/j.chemolab.2026.105628","url":null,"abstract":"<div><div>Ash content is a key quality index in coal preparation. To overcome the time-consuming, labor-intensive nature of offline assays, the interference sensitivity of online measurements, and the surface-only information of vision/spectroscopy, this study proposes a multimodal deep-learning framework for fast and accurate raw-coal ash prediction. First, a CT-based data acquisition system was designed to synchronously collect CT image slices of raw-coal particles and particle-density information. Through quantitative analysis of the association between density and ash content, the decisive role of density – as a key indicator of coal properties – in ash prediction is revealed. By further mining the frequency-domain and spatial features of coal CT images, and based on these findings, a multimodal approach is formulated that fuses CT features with density. The model adopts a three-branch architecture: an improved EfficientNet-B0 branch learns spatial cues, an AshFormer branch captures frequency patterns related to mineral distribution and microstructural discontinuities, and a multilayer perceptron encodes density. Cross-modal attention achieves deep fusion and complementarity across modalities, and a KAN-based regression head outputs ash content. On industrial data, the proposed method attains MAPE = 0.0468, RMSE = 0.0573, and <span><math><mrow><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup><mo>=</mo><mn>97</mn><mo>.</mo><mn>0</mn><mtext>%</mtext></mrow></math></span>, outperforming single-modality image models (<span><math><mrow><mi>Δ</mi><mi>M</mi><mi>A</mi><mi>P</mi><mi>E</mi><mo>=</mo><mo>−</mo><mn>0</mn><mo>.</mo><mn>0096</mn></mrow></math></span>, <span><math><mrow><mi>Δ</mi><mi>R</mi><mi>M</mi><mi>S</mi><mi>E</mi><mo>=</mo><mo>−</mo><mn>0</mn><mo>.</mo><mn>0413</mn></mrow></math></span>, <span><math><mrow><mi>Δ</mi><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup><mo>=</mo><mo>+</mo><mn>4</mn><mo>.</mo><mn>48</mn></mrow></math></span> percentage points). These results demonstrate the advantage of multimodal fusion in improving the accuracy and generalization of coal-quality analysis and provide a new approach for rapid analysis under small-sample conditions.</div></div>","PeriodicalId":9774,"journal":{"name":"Chemometrics and Intelligent Laboratory Systems","volume":"269 ","pages":"Article 105628"},"PeriodicalIF":3.8,"publicationDate":"2026-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145920782","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0