Journal of Food Engineering最新文献

{"title":"AI-based framework to detect azo-dye adulteration in red chilli powder: ChoLevy-R optimiser for convolution neural network optimisation and validated using FTIR","authors":"Dilpreet Singh Brar ,&nbsp;Birmohan Singh ,&nbsp;Vikas Nanda","doi":"10.1016/j.jfoodeng.2026.113000","DOIUrl":"10.1016/j.jfoodeng.2026.113000","url":null,"abstract":"<div><div>The aim of this study is to evaluate the effect of Sudan(IV) adulteration on the antioxidant activity of red chilli powder (RCP) and to develop a multistage AI-based detection framework for identifying such adulteration within the industrial processing and supply chain of RCP. This paper proposed a novel optimiser, ChoLevy-R, that integrates chaotic dynamics with Lévy flight perturbations on the RMSprop update rule to improve the training stability of DenseNet-169 for identifying Sudan(IV) adulteration in red chilli powder (RCP). The DenseNet-169 with ChoLevy-R delivered a validation accuracy of 95.78 % and stabilised the training and validation processes. The further Grad-CAM heatmap provided insight into the model's decision-making, assigning a confidence score of 1.00 for all classes. The proposed AI framework was validated against FTIR spectroscopy with Support Vector Machine (SVM) classification. Moreover, PCA loading highlighted significant wavenumbers at 1300-1400 cm⁻¹ (–N=N–) and 1400-1600 cm⁻¹ (C=C stretches) linked to Sudan(IV) structural components. These influential wavenumbers enabled the SVM to achieve an adulteration detection accuracy of 96.55 %. Overall, the ChoLevy-R-optimised DenseNet-169 architecture, validated by FTIR spectroscopy, provides a dependable, interpretable, and scalable approach for real-time adulteration detection in RCP, facilitating robust food quality analyses in the food processing industry.</div></div>","PeriodicalId":359,"journal":{"name":"Journal of Food Engineering","volume":"413 ","pages":"Article 113000"},"PeriodicalIF":5.8,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146057406","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Simulation study of Escherichia coli electroporation induced by pulsed magnetic field","authors":"Fei Guo,&nbsp;Hao Yuan,&nbsp;Guilin Dan","doi":"10.1016/j.jfoodeng.2026.113013","DOIUrl":"10.1016/j.jfoodeng.2026.113013","url":null,"abstract":"<div><div>This paper introduces an innovative simulation model designed to explore, for the first time, the phenomenon of electroporation (EP) in <em>Escherichia coli</em> (<em>E. coli</em>) under the influence of a pulsed magnetic field (PMF). Initially, to ensure the viability and precision of the simulation model, a comparative validation was carried out. Compared with analytical model, simulation model yields more accurate results. Subsequently, the study investigates how two pivotal geometric parameters (cell radius and membrane thickness) impact the transmembrane voltage (TMV). The findings reveal that increasing the cell radius from 0.25 μm to 0.5 μm results in an approximate 0.47V rise in the maximum TMV, while increasing the cell membrane thickness from 24 nm to 29 nm leads to an approximate 0.16 V rise in the maximum TMV. Moreover, the maximum changes in pore density and conductivity have increased by approximately two orders of magnitude. Finally, to effectively validating its predictive capability regarding the likelihood of EP occurrence, the simulation model was expanded and compared against experimental data. In summary, this paper is dedicated to elucidating the complex dynamic behaviors and underlying mechanisms of irregular cells during the EP process, aiming to provide a solid theoretical foundation and crucial guidance for the potential application of PMF technology in the future food industry.</div></div>","PeriodicalId":359,"journal":{"name":"Journal of Food Engineering","volume":"413 ","pages":"Article 113013"},"PeriodicalIF":5.8,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146185768","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Coaxial 3D printing of surimi-based composite fruit–vegetable gels","authors":"Qiang Tong ,&nbsp;Junhao Yang ,&nbsp;Peng Wu ,&nbsp;Yanwu Cao ,&nbsp;Xiuping Dong ,&nbsp;Guibing Pang","doi":"10.1016/j.jfoodeng.2026.112970","DOIUrl":"10.1016/j.jfoodeng.2026.112970","url":null,"abstract":"<div><div>Food 3D printing enables precise structuring of food materials and shows strong potential for personalized nutrition. Yet most studies focus on single-material systems, with limited application of composite and multi-nutrient co-printing. In this study, a coaxial extrusion-based 3D printing system was established using surimi as the outer phase and spinach gel as the inner phase. The rheological properties, moisture distribution,and molding performance of various fruit and vegetable gels (spinach, pumpkin, strawberry, yam, and mango) were systematically analyzed. The results showed that the spinach gel exhibited the highest G′ value and a moderate tan δ (&lt;0.7). Meanwhile, its T₂ peak (40–50 ms), together with high hardness and a pronounced elastic response obtained from single compression tests, indicated a well-balanced compromise between structural support and flowability.COMSOL simulations combined with printing experiments showed that, at an outer-phase printing speed of 0.05 mm/s and an inner-phase speed of 0.06 mm/s, the system can stably fabricate geometric models such as cubes, triangles, and pyramids, further confirming the excellent forming performance of the coaxial extrusion system. Measurements under different cooking methods indicated that the coaxial structure provides a clear protective effect on the nutrients in the inner phase. Taken together, the overall objective of this study was to develop a coaxial extrusion-based 3D-printed food system that achieved high shape fidelity while effectively protecting inner-phase nutrients.</div></div>","PeriodicalId":359,"journal":{"name":"Journal of Food Engineering","volume":"412 ","pages":"Article 112970"},"PeriodicalIF":5.8,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145924287","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Bayesian Gaussian Process Regression soft sensor for industrial sugar crystallization monitoring","authors":"Humberto Morales ,&nbsp;Estefania Aguirre-Zapata ,&nbsp;Xudong Shi ,&nbsp;Fernando di Sciascio ,&nbsp;Adriana N. Amicarelli","doi":"10.1016/j.jfoodeng.2026.112990","DOIUrl":"10.1016/j.jfoodeng.2026.112990","url":null,"abstract":"<div><div>Supersaturation is the primary variable in sugar crystallization, as it governs sucrose crystal growth and product quality, whereas the purity of the mother liquor is a key indicator of process efficiency. Both variables are difficult to measure online due to technological limitations. This study presents data-driven soft sensors based on Bayesian Gaussian Process Regression (BGPR) for predicting supersaturation and purity in industrial crystallization. A novel covariance matrix was introduced through an extended kernel design, enhancing the flexibility of the BGPR framework. Beyond accurate point predictions, the BGPR framework inherently provides calibrated uncertainty quantification, addressing an issue rarely discussed in the crystallization literature. The proposed soft sensors were trained and validated with real data from an industrial vacuum pan, achieving relative errors below 4% for both variables. The soft sensor requires only three routinely measured variables: temperature, magma brix, and mother liquor brix, which represents a practical advantage since no additional instrumentation is needed. For supersaturation, the uncertainty analysis yielded a Prediction Interval Coverage Probability (PICP) of 93.1%, narrow prediction intervals (MPIW = 0.1024), and a favorable CRPS of 0.0317, confirming robustness. The BGPR soft sensor was compared with five widely used approaches in soft-sensor design: Artificial Neural Networks (ANN), Long Short-Term Memory networks (LSTM), Support Vector Regression (SVR), Random Forest (RF), and Gradient Boosting Regression (GBR). Results showed that BGPR achieves competitive predictive accuracy while additionally providing reliable information on estimation uncertainty. These findings demonstrate the applicability of BGPR-based soft sensors for real-time monitoring of industrial sugar crystallization.</div></div>","PeriodicalId":359,"journal":{"name":"Journal of Food Engineering","volume":"412 ","pages":"Article 112990"},"PeriodicalIF":5.8,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146035310","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimization, simulation, and application of stamp-based fabrication method of paper-based microfluidics device for food safety analysis with smartphone app-assisted detection","authors":"K. R. Jolvis Pou ,&nbsp;Shervin Foroughi ,&nbsp;Dhilippan M. Panneerselvam ,&nbsp;Muthukumaran Packirisamy ,&nbsp;Vijaya Raghavan","doi":"10.1016/j.jfoodeng.2026.113005","DOIUrl":"10.1016/j.jfoodeng.2026.113005","url":null,"abstract":"<div><div>Over the last decade, microfluidic paper-based analytical devices (μPADs) have emerged as innovative platforms due to their cost-effectiveness, compatibility, availability, user-friendliness, and disposability. Wax printing is one of the most widely used techniques for the fabrication of μPADs. This study aims to optimize wax-printing fabrication parameters, simulate pressure and fluid dynamics, and apply the optimized conditions to create various microfluidic patterns and develop a pH indicator for food safety analysis. It also focuses on developing a smartphone app to automate the results of chromogenic reactions. The influences of process parameters, namely molten wax temperature (100–160 °C), applied pressure (0.1–6 g/mm²), stamp width (0.5–2 mm), and holding time (1–10 s) on the printing quality and wax spreading were studied. A smartphone app was developed using the Thunkable workspace, with Imagga integrated as the application programming interface platform. The results indicated that Whatman grade 4 exhibited the highest fluid velocity among the tested paper types, which is suggested as a suitable substrate for milk safety analysis. The optimal fabrication conditions for achieving high-quality wax printing with minimal wax spreading were determined to be at 115 °C, 1.575 g/mm², 0.875 mm, and 3.25 s for temperature, pressure, stamp width, and holding time, respectively. Using these optimized conditions, a paper-based pH sticker was developed to detect pH levels ranging from 1 to 10 for milk safety analysis. The color output ranged from red at pH 1 to blue at pH 10, with a greenish hue observed around the neutral pH range. These results indicated the reliability and adaptability of the device for on-site detection of food safety.</div></div>","PeriodicalId":359,"journal":{"name":"Journal of Food Engineering","volume":"413 ","pages":"Article 113005"},"PeriodicalIF":5.8,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146185767","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"In situ rheological monitoring during in vitro gastric digestion: Linking viscosity dynamics with lipid bioaccessibility in oil-in-water emulsions","authors":"Carolina Quezada ,&nbsp;Andrés Arrué ,&nbsp;Camila Mella ,&nbsp;Karla Pérez ,&nbsp;Rommy N. Zúñiga ,&nbsp;Elizabeth Troncoso","doi":"10.1016/j.jfoodeng.2026.113020","DOIUrl":"10.1016/j.jfoodeng.2026.113020","url":null,"abstract":"<div><div>Understanding how food viscosity evolves during digestion is essential to predict nutrient release, gastric emptying, and the functional performance of food matrices. However, conventional <em>ex situ</em> rheological measurements require enzymatic activity to be halted, potentially altering sample properties and limiting physiological relevance. In this study, we developed and tested a novel <em>in vitro</em> gastric digestion system integrated with <em>in situ</em> rheological monitoring (iGD-isRM) to continuously track apparent viscosity during digestion of oil-in-water (O/W) emulsions. Emulsions stabilized with either Quillaja saponin or soy lecithin and thickened with modified starch (oil-to-surfactant, O:S, ratios of 5:1 and 10:1) were characterized under simulated gastric conditions, including dynamic acidity adjustment, secretion-driven dilution, and controlled gastric emptying. Results showed a sharp decrease in apparent viscosity within the first 15 min, driven primarily by dilution by gastric secretions rather than enzymatic hydrolysis. Saponin-stabilized emulsions exhibited higher initial viscosity (2495-2771 mPa s) but dropped more rapidly to near-zero values during gastric digestion. Their smaller droplet sizes (0.19-0.34 μm) also contributed to greater free fatty acid (FFA) release during the intestinal phase (40.6-44.4%). In contrast, lecithin-stabilized emulsions displayed lower initial viscosity (551-1146 mPa s), a slower decrease toward near-zero values, larger droplet sizes (0.51-0.94 μm), and consequently lower FFA release (36.6-37.5%). Additionally, lower viscosity and surfactant concentration (O:S = 10:1) enhanced enzyme diffusion and improved lipid bioaccessibility. The iGD-isRM system enabled reliable real-time rheological measurements and revealed a clear relationship between chyme viscosity dynamics, structural disintegration, and intestinal lipolysis. This study demonstrates the potential of <em>in situ</em> rheological monitoring to improve our mechanistic understanding of digestion processes and to guide the rational design of structured food matrices for targeted nutrient delivery.</div></div>","PeriodicalId":359,"journal":{"name":"Journal of Food Engineering","volume":"413 ","pages":"Article 113020"},"PeriodicalIF":5.8,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146185784","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Synergistic effects of pulsed electric field and ultrasound on microstructure and bioactive properties of fresh and old ginger juice","authors":"Zhong Han ,&nbsp;Wenhao Xiao ,&nbsp;Ji Ma ,&nbsp;Muhammad Mohsin Raza ,&nbsp;Xin-An Zeng ,&nbsp;Muhammad Faisal Manzoor ,&nbsp;Muhammad Talha Afraz","doi":"10.1016/j.jfoodeng.2026.112992","DOIUrl":"10.1016/j.jfoodeng.2026.112992","url":null,"abstract":"<div><div>Ginger (<em>Zingiber officinale</em>) contains abundant bioactive compounds such as gingerols, shogaols, and phenolic antioxidants, which contribute to its nutritional and functional importance. However, efficient extraction remains limited due to incomplete cell wall disruption under single-field treatments. The combination of pulsed electric field (PEF) and ultrasound (US) offers a synergistic approach, where electroporation and acoustic cavitation collectively enhance cell permeability and mass transfer. To address this, the present study directly compares the combined PEF–US treatment on fresh versus old ginger for the first time. We developed a PEF–US coupling system (2.5 kV/cm and 450 W) and applied the combined treatment to both fresh and old ginger to evaluate its effects on juice yield and bioactive compound release. Under the optimized PEF–US conditions, both fresh and old ginger improved, with the strongest response in old ginger. Specifically, OG-2.5–450W produced 92.24 ± 0.88 % juice yield and 794.02 ± 1.15 μg GAE/mL total phenolic content (<em>p &lt; 0.05</em>), together with about 79.9 % DPPH^• radical inhibition and 2315.05 ± 43.01 μg/g 6-gingerol (<em>p &lt; 0.05</em>), exceeding the single-treatment and fresh-ginger groups. Structural and chemical changes supporting these outcomes were evidenced by FTIR, microscopy, and a high disintegration index (Z = 0.71 ± 0.02). Overall, the combined PEF–US treatment substantially enhances extraction efficiency and antioxidant-related bioactivity, particularly in old ginger. This innovative extraction strategy offers broader industrial relevance, providing a scalable and efficient approach for producing high–value ginger extracts suitable for functional foods, beverages, nutraceuticals, and natural health product formulations.</div></div>","PeriodicalId":359,"journal":{"name":"Journal of Food Engineering","volume":"413 ","pages":"Article 112992"},"PeriodicalIF":5.8,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146057404","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine learning-based predictive modelling of wine fermentation dynamics using temperature control data","authors":"Angel Sanchez-Roca ,&nbsp;Emilio Jiménez-Macías ,&nbsp;Juan-Ignacio Latorre-Biel ,&nbsp;Juan Carlos Saenz-Díez ,&nbsp;Mercedes Pérez de la Parte ,&nbsp;Julio Blanco-Fernández","doi":"10.1016/j.jfoodeng.2026.112969","DOIUrl":"10.1016/j.jfoodeng.2026.112969","url":null,"abstract":"<div><div>Wine fermentation is a complex exothermic biochemical process in which the glucose and fructose presents in the must is transformed into ethanol and carbon dioxide, owing to the action of yeasts. This experimental study evaluates the feasibility of using machine learning as a tool to model the dynamic behavior of wine fermentation, using characteristics extracted from temperature control data as the main variable of the process. The measurements were made in several fermentation tanks at a two-winery producing wine. The temperature time signals from industrial-scale fermentation tanks were acquired with a sampling period of 15 min and stored for later processing. Relevant thermal characteristics, such as temperature gradients, rates of change, and control system response patterns, were extracted from the time series obtained. Various supervised learning algorithms—including Linear Regression Models, Support Vector Machines, Regression Trees, Ensemble of Trees, and Gaussian Process Regression —were trained and validated to predict the dynamics of must density evolution. To validate the results, measurements obtained manually by the winemaker were compiled. The performance of the models was evaluated using standard statistical metrics demonstrating high predictive accuracy and robustness across different fermentation batches. The results of the analyses confirm a high linear correlation between the density obtained from the machine learning-based model and the densities obtained at the laboratory level, with coefficients of determination greater than 95 %. The proposed methodology is a useful tool for advancing predictive modeling of oenological systems, contributing to the development of intelligent control strategies and aiding decision-making in wineries.</div></div>","PeriodicalId":359,"journal":{"name":"Journal of Food Engineering","volume":"412 ","pages":"Article 112969"},"PeriodicalIF":5.8,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145975539","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Effect of 12-HSA concentration on the physicochemical properties, stability, and curcumin bioavailability of ultrasound-assisted enzymatically glycosylated casein-stabilized oleogel-structured emulsions","authors":"Yan Li ,&nbsp;Wanlinhao Wei ,&nbsp;Yingying Sui ,&nbsp;Olayemi Eyituoyo Dudu ,&nbsp;Wentao Liu ,&nbsp;Mingrui Zhang ,&nbsp;Jinju Cheng ,&nbsp;Yujun Jiang","doi":"10.1016/j.jfoodeng.2026.112971","DOIUrl":"10.1016/j.jfoodeng.2026.112971","url":null,"abstract":"<div><div>To investigate the effects of glycosylated covalent complexes and 12-hydroxystearic acid (12-HSA) on the properties of oleogel-structured high internal phase emulsions (OGHEs), this study prepared OGHEs with varying 12-HSA concentrations (0 %, 1 %, 3 %, 5 %, and 7 %) using 1.5 % sodium caseinate (SC) or conjugates of SC and carboxymethyl chitosan (SC-CMCS) as emulsifiers. A systematic evaluation was conducted on their droplet size, zeta potential, interfacial adsorbed protein content, rheological properties, stability, curcumin encapsulation efficiency, and in vitro digestion performance. Results showed that SC-CMCS-stabilized OGHEs exhibited significantly superior apparent viscosity and elastic modulus (with G′ consistently greater than G'') compared to SC-stabilized OGHEs. The elastic modulus (G′) reached its maximum at 7 % 12-HSA, accompanied by a smaller initial droplet size (8.02 ± 0.17 μm) and more uniform distribution. Confocal laser scanning microscopy (CLSM) revealed a dual-stabilization structure consisting of oil droplets and a fibrous network, while differential scanning calorimetry (DSC) confirmed enhanced thermodynamic stability. Meanwhile, SC-CMCS-stabilized OGHEs with high 12-HSA concentrations demonstrated markedly improved thermal stability, freeze-thaw stability, and lipid oxidation stability. After 28 d of storage, their particle size increased only to 11.88 ± 0.79 μm, and the curcumin encapsulation efficiency significantly rose to 95.85 ± 0.31 %, with a retention rate of 96.32 ± 0.38 % after 12 d of storage. These results indicate that the synergistic interaction between SC-CMCS and 12-HSA comprehensively enhances the functionality of OGHEs, providing theoretical support for the development of carriers for active ingredients in functional foods.</div></div>","PeriodicalId":359,"journal":{"name":"Journal of Food Engineering","volume":"412 ","pages":"Article 112971"},"PeriodicalIF":5.8,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145924274","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development of a high-accuracy multilayer perceptron-based soft sensor for real-time monitoring of supersaturation and dry substance content in vacuum pan crystallization","authors":"S.M. Petrov ,&nbsp;D.V. Arapov ,&nbsp;V.A. Kuritsyn ,&nbsp;N.M. Podgornova","doi":"10.1016/j.jfoodeng.2026.112966","DOIUrl":"10.1016/j.jfoodeng.2026.112966","url":null,"abstract":"<div><div>Sucrose crystallization is a key stage in sugar production, where control of the supersaturation coefficient (<em>SS</em>) and dry substance content (<em>DS</em>) is critical for product quality and energy efficiency. Direct online monitoring of these parameters remains challenging. This paper proposes a soft sensor based on a five-layer multilayer perceptron (MLP) for simultaneous prediction of <em>SS</em> and <em>DS</em>. The model utilizes three easily measurable input parameters: boiling point temperature of pure water, boiling point elevation (BPE), and the estimated purity of the intercrystalline solution. Training was performed using the RMSProp algorithm on an extensive dataset of experimental data. The developed model demonstrated superior accuracy, with a mean relative error of 1.0 % for <em>SS</em> and 0.49 % for <em>DS</em>, which is 2.1–7.0 times more accurate than state-of-the-art alternatives. An architecture for an automatic control system integrating the developed soft sensor for real-time closed-loop control is presented. Our results show that the proposed approach is a highly effective solution for automating and optimizing the massecuite boiling process.</div></div>","PeriodicalId":359,"journal":{"name":"Journal of Food Engineering","volume":"412 ","pages":"Article 112966"},"PeriodicalIF":5.8,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145924288","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}