Sugar Tech最新文献

Sugar crystallization is a tightly controlled unit operation in which variations in supersaturation, temperature, and impurity load can directly affect Brix, purity, color, and crystal size. However, data-driven soft sensors often face limited labeled data, measurement noise, missing observations, and poor generalization under changing operating conditions. This study presents a synthetic digital-twin benchmark for vacuum pan crystallization and a physics-guided machine-learning framework based on gradient-boosting decision trees for simultaneous prediction of multiple quality attributes. The benchmark generates batch trajectories using simplified mass–energy balances and crystallization-rate proxies, while incorporating industrially relevant sensor noise, drift, missingness, actuator lag, feed-quality disturbances, and operating-regime shifts. Windowed process histories are transformed into process-informed descriptors and used to train gradient-boosting models for Brix, purity, color, and crystal-size prediction. Physical plausibility was assessed through process-informed features, monotonicity checks, and residual-based scoring, while robustness was evaluated using in-distribution (Test-ID) and out-of-distribution (Test-OOD) scenarios and conformal prediction intervals. The proposed models achieved strong nominal performance for Brix, purity, and crystal size, with Test-ID RMSE values of 0.440, 0.810, and 18.502, respectively. Under operating shifts, Brix and purity remained comparatively stable, whereas color and crystal size showed greater performance degradation, indicating higher sensitivity to regime changes. Conformal analysis provided informative uncertainty estimates under nominal conditions but revealed reduced coverage under Test-OOD conditions. Overall, the results demonstrate the potential of gradient-boosting soft sensors for crystallization monitoring while highlighting the need for improved uncertainty quantification and distribution-shift resilience for reliable industrial deployment.

Starch is currently recognized as a critical quality parameter in raw sugar production. Despite its importance, few studies have systematically tracked the distribution and fate of starch throughout the sugar manufacturing process. This study provides a comprehensive evaluation of starch behavior across nine process streams, from juice extraction to its final incorporation within raw sugar crystals, including several intermediate streams that have received limited attention in previous research. Starch concentration and mass flow were determined in the extraction, purification, and crystallization stages. In addition, relative errors were calculated at each stage to assess the quality of the analytical measurements and the consistency of the material balances. The results showed that diluted and filtered juices contained high starch concentrations due to the characteristics of their generation and handling. During purification, approximately 61% of the starch was removed, whereas 24% was retained in the sugar product. Furthermore, 70% of the starch present in the final product was located within the core of the raw sugar crystal. The filtered juice recirculated 23.8% of the starch removed during clarification back to the alkalization tank, highlighting an important pathway for starch redistribution within the process. Based on these findings, a starch distribution map was developed to describe the fate of starch throughout sugar manufacturing and to identify critical stages for its control. The proposed map improves the understanding of starch dynamics during sugar processing and provides a basis for implementing more effective quality control and enzymatic treatment strategies.

The paper analyses BRICS sugar production scenario, intra-BRICS sugar trade, and the competitive landscape of sugar trade of each member country. BRICS accounts for 55.44% of world sugar production, 90% of which is produced from sugarcane. Intra-BRICS sugar trade is worth 17.18 billion USD representing 41.18% of world sugar trade. The member nations prefer intra-BRICS sugar imports while they are less dependent on intra-BRICS exports. Intra-BRICS sugar exports as per cent of country’s total sugar exports were quite varied from India and Indonesia (6–8%), China (17%) to Brazil (38%). India, Egypt, Ethiopia, and Iran depend more on intra-BRICS sugar imports, up to 95% of their total quantum compared to South Africa (15%). Many BRICS members are persistently exporting conventional products while a few have diversified their exports. Brazil, Egypt, India, and South Africa have the competitiveness in the export of “sugar, molasses and jaggery” product group, while Iran, Egypt, and South Africa have competitive edge in the export of sugar confectionery. The average tariff levels of all BRICS members are quite varied, high for some members, and biased also towards some countries. Brazil faces the highest, and Ethiopia benefits from the lowest Indian tariff. South Africa and UAE apply the highest tariffs on Brazil products. Though sugar trade facilitating business environment in member countries has improved over the years, India needs to develop a long-term strategy for strengthening intra-BRICS investment and trade relations. India also needs to be proactive for enhancing policy-level cooperation amongst members for diversification of sugar exports and generating more conducive and facilitating intra-BRICS sugar production-cum-trade environment.

Early yield estimation is essential for agricultural planning, localized management and decision-making in commercial areas, especially in view of the high spatial variability of sugarcane fields. However, traditional estimation methods are generally destructive, costly, and inefficient in accounting for this variability. In this context, the integration of remote sensing data, soil attributes and machine learning techniques presents itself as a promising alternative for predicting sugarcane productivity. The objective of this study was to apply predictive models of productivity at a sub-field scale, integrating orbital data, structural information derived from LiDAR (Light Detection and Ranging) and mineralogical attributes of the clay fraction. The study was conducted in a commercial area in the state of São Paulo. The actual productivity data were obtained by sensors embedded in a sugarcane harvester, georeferenced by GNSS with RTK correction. Multispectral images of the constellation Planet Labs were used, from which spectral bands, vegetation indices and texture attributes based on the Gray Level Co-occurrence Matrix (GLCM) were extracted. The height of the plants was obtained by airborne LiDAR surveys, while the soil mineralogy was characterized by magnetic susceptibility measurements, converted into a representative index of the clay fraction. The Random Forest, XGBoost, KNN and Multiple Linear Regression models were evaluated. The results indicated that the XGBoost model presented the best performance among the models tested after selecting the variables, with RMSE and MAPE values of 5.53 t ha⁻¹ and 5.79%, highlighting the texture attributes derived from GLCM as the most important variables for yield prediction, surpassing plant height and soil mineralogy data.

Stevia rebaudiana is a source of low-calorie, non-nutritive steviol glycosides (SGs). Among all SGs, Rebaudioside A, Rebaudioside D, and Rebaudioside M are in higher commercial demand due to their less bitter aftertaste. While Rebaudioside A is one of the predominant SG, RebD and RebM occur at relatively low concentration. Developing stevia varieties with enhanced accumulation of Reb A, D, and Mare among the current breeding challenges. Additionally, limited information exists on the regulatory mechanisms underlying the multisubstrate functionality of key UDP-glycosyltransferases (UGTs) that synthesize ~ 60 different SGs. Transcriptome-derived SNPs, ascertained from actively transcribed regions, are particularly relevant for gene function and regulation studies compared to non-genic SNPs. In this study, transcriptome-derived single nucleotide polymorphisms (SNPs) were identified using RNA-seq data from nine contrasting genotypes with varying concentrations of Reb A, D, and M. A total of 8, 40,332 high-quality SNPs were identified, and categorized into 1, 82,802 missense categories and 3,960 nonsense categories. Genomic localization showed the maximally in intergenic regions (3, 59,023), followed by genic (3, 41,137) and intronic (1, 97,976) regions. Functional enrichment analysis revealed that DEGs associated with key traits, including growth biomass, secondary metabolism, and abiotic stress-related categories, were highly enriched. Furthermore, 501 potential missense variations were identified associated with the SG biosynthesis pathway. Molecular docking revealed six potential variants in the UGT76G1 gene that showed higher substrate selectivity toward Reb A (ile234ser, phe22leu, and ser357thr) and RebD (ile16val, pro240ser, and lys39glu). These targeted variants can accelerate the genetic improvement program for stevia varieties with elevated accumulation of RebA and RebD. The SNP resources can also support the development of high-density genotyping platforms and GWAS studies.

Organic fertilizers can provide economic and ecological benefits for nutrient management in soil–plant systems. However, the long-term effects of chicken manure application on rhizosphere microbial ecology and sugar beet yield remain unclear. Here, we conducted a six-year field experiment with CM application in a northern black soil area of China to analyze the long-term effects of CM on soil physicochemical and microbiological properties, as well as on sugar beet yield. Results indicated that long-term CM application increased sugar beet yield by 33.83%, soil organic matter (SOM) by 37.80%, and the soil quality index by 10.90% compared with no CM application. Continuous CM application also increased bacterial OTU richness and community diversity, enhanced predicted metabolic and membrane transport functions, and increased the complexity of bacterial co-occurrence networks in the rhizosphere soil. The use of CM also reduced the relative abundance of fungal taxa assigned as plant pathogens by 24.68%, but with a concomitant 4.35% increase in the relative abundance of rhizosphere fungal taxa assigned to animal pathogens in the rhizosphere fungi. In conclusion, long-term CM input enhanced the overall quality of black soil, optimized the sugar beet rhizosphere ecological environment, and mitigated surface source pollution from organic waste in animal husbandry. This study provides a basis for improving circular nutrient use, promoting on-site manure recycling, and supporting sustainable agricultural production.

Field experiment was conducted during the spring season 2021–2022 and 2022–2023 to study the effect of band application of herbicides in combination with mulching or hoeing on weed control in sugarcane. The experiment was investigated in randomized complete block design with twelve weed control treatments which comprised of weedy check, mulching @ 6 t ha⁻¹, hand-hoeings at 45, 75 and 105 DAP (days after planting), metribuzin (blanket) 1400 g ha⁻¹ as pre-emergence (PE), metribuzin (band) 560 g ha⁻¹ followed by (fb) mulching @ 6 t ha⁻¹, metribuzin (band) 560 g ha⁻¹ fb three hand-hoeings, sulfentrazone + clomazone (blanket) 700 + 750 g ha⁻¹ as PE, sulfentrazone + clomazone (band) 280 + 300 g ha⁻¹ fb mulching @ 6 t ha⁻¹, sulfentrazone + clomazone (band) 280 + 300 g ha⁻¹ fb three hand-hoeings, metribuzin 1050 g ha⁻¹ + tembotrione 115.5 g ha⁻¹ (blanket) as post-emergence (POE), mulching @ 6 t ha⁻¹ fb metribuzin 420 g ha⁻¹ + tembotrione 46.5 g ha⁻¹ (band), metribuzin 420 g ha⁻¹ + tembotrione 46.5 g ha⁻¹ (band) with three hand-hoeings. Dactyloctenium aegyptium, Echinochloa colona, Cyperus rotundus, Digera arvensis and Convolvulus arvensis were the major weeds in the experimental crop. Pre-emergence band application of sulfentrazone + clomazone 280 + 300 g ha⁻¹ fb mulching @ 6 t ha⁻¹ was highly effective against the grasses, the only sedge and broad-leaf weeds followed by metribuzin 560 g ha⁻¹ with mulching @ 6 t ha⁻¹. Cane yield was the highest (92.4 t ha⁻¹) in sulfentrazone + clomazone (band) 280 + 300 g ha⁻¹ fb mulching @ 6 t ha⁻¹ with the highest net returns and B:C ratio. Higher cane yield in band application of sulfentrazone + clomazone 280 + 300 g ha⁻¹ fb mulching might be attributed to higher tiller population (137.0 thousand per ha), cane length, cane girth, single cane weight (978.8 g) and number of millable canes (96.4 thousand per ha) due to better weed control.

Sugarcane in Yunnan, China, is predominantly cultivated in high-elevation, rainfed systems where water availability mainly limits productivity. Soil potassium (K) plays a key role in drought tolerance, and optimized K fertilization may help mitigate water limitations. However, systematic evaluation of elite genotypes across different potassium fertilizer levels in Yunnan's conditions is lacking. This study evaluated 55 sugarcane genotypes under three K fertilization levels: K0 (0), K1 (180), and K2 (360) kg K₂O ha⁻¹. A composite Z-score index, based on plant height, stalk diameter, and field Brix at 11 months, was used to select 19 representative genotypes for biomass and tissue K analyses. Moderate K (180 kg K₂O ha⁻¹) supported vegetative growth with no gain from higher rates; the highest Brix under K0 was due to biomass dilution effect, not to increased sugar synthesis per plant. Based on the results of plant biomass and tissue K accumulation indices, four distinct genotypic response categories were identified: constitutively K-efficient (YZ151679, FY798, FL41), K-inducible (GL03182, GL05136), K-dependent efficient (YT63-237), and K-utilization efficient (MT70-611). Post-harvest soil analysis revealed that potassium fertilization increased soil pH from 5.11 to 6.65. Soil organic matter exhibited strong positive correlations with available K (r = 0.874) and total N (r = 0.813). Furthermore, total N was positively correlated with available K (r = 0.945), indicating tightly coupled K-N dynamics. These findings provide the first systematic framework for selecting K-efficient parental stocks: constitutively efficient and utilization-efficient genotypes for low-input systems, and K-inducible or K-dependent efficient genotypes for fertilized systems.

The present study was undertaken to develop and optimize a jaggery-based chocolate enriched with sprouted finger millet flour and to assess its nutritional and textural characteristics. Response surface methodology (RSM) was applied using Design-Expert software (ver. 11.1.0.1). The independent variables selected were cocoa butter replacement with coconut oil (20–30% w/w), conching time (120–180 min), and sprouted finger millet flour incorporation (20–40% w/w). A total of seventeen experiments were performed, and the responses were modelled through multiple regression analysis. The optimized formulation, exhibiting maximum overall desirability, comprised of 22% (w/w) cocoa butter replacement with coconut oil, 150 min of conching, and a 39% (w/w) incorporation of sprouted finger millet flour. The optimized product contained 5.16% moisture, 2.66% ash, 23.60% crude fat, and 9.87% crude protein. Textural analysis revealed hardness of 305.6 N, adhesiveness of 2.14 N·s, cohesiveness of 0.101, and chewiness of 4.07 N. Sensory evaluation indicated high acceptability, with mouthfeel and overall acceptability scores of 8.12 and 7.80, respectively. Model validation demonstrated no significant differences between predicted and experimental values, confirming the adequacy and reliability of the developed models.

This study was conducted to determine and optimize energy indices and greenhouse gas emissions, as well as to perform an economic analysis of sugar beet production using Data Envelopment Analysis (DEA). The required data were collected through interviews with farmers and agricultural experts in Eyvan County, Ilam Province of Iran, as well as via structured questionnaires. The results indicated that the total energy input and output were 45,211.99 and 1,093,411.2 MJ ha⁻¹, respectively. Regression coefficients revealed that human labor, machinery, phosphorus and potassium fertilizers, and also manure had a positive effect on crop yield, whereas electricity, nitrogen fertilizer, and diesel fuel had a negative impact. The Greenhouse Gas (GHG) emission analysis indicated that, under conventional practices, nitrogen fertilizer and diesel fuel contributed 620.19 and 606.24 kg CO₂-eq, respectively; after optimization, these values were reduced to 561.54 and 438.74 kg CO₂-eq. DEA of energy efficiency showed that, under the Constant Returns to Scale (CRS) model, 32% of units were fully efficient, while under the Variable Returns to Scale (VRS) model, 72% of units achieved full efficiency, with the remaining units considered inefficient to varying degrees. The optimization results of input use indicated that, in the VRS model, nitrogen fertilizer, manure, and agricultural machinery accounted for the highest energy-saving potential, with 14.97%, 14.73%, and 10.47%, respectively. In the CRS model, agricultural machinery, diesel fuel, and manure represented the greatest potential for energy conservation, with 38.55%, 27.63%, and 22.38%, respectively. The economic analysis showed that, compared to conventional practices, variable costs in the optimized scenario decreased by 8.08%, and total production costs decreased by 6.39%. As a result, gross revenue increased by 7.76%, while net revenue (and consequently net return per hectare) increased by 10.38% compared to the conventional scenario.