Article(id=1322631260816658821, tenantId=1146029695717560320, journalId=1302319053441957962, issueId=1322631228000424183, articleNumber=null, orderNo=null, doi=10.7501/j.issn.0253-2670.2026.18.020, pmid=null, cstr=null, oa=null, hot=0, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=null, receivedDate=1774886400000, receivedDateStr=2026-03-31, revisedDate=null, revisedDateStr=null, acceptedDate=null, acceptedDateStr=null, onlineDate=1793269465878, onlineDateStr=2026-10-29, pubDate=null, pubDateStr=null, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1793269465878, onlineIssueDateStr=2026-10-29, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1793269465878, creator=13701087609, updateTime=1793269465878, updator=13701087609, issue=Issue{id=1322631228000424183, tenantId=1146029695717560320, journalId=1302319053441957962, year='2026', volume='57', issue='18', pageStart='7073', pageEnd='7500', issueExtLink='null', onlineDate='null', pubDate='1790524800000', pubDateStr='2026-09-28', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=null, createTime=1793269458054, creator='13701087609', updateTime=1793269548245, updator='13701087609', preIssue=null, nextIssue=null, articleTotal=null, ext={EN=IssueExt(id=1322631606339228314, tenantId=1146029695717560320, journalId=1302319053441957962, issueId=1322631228000424183, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1322631606339228315, tenantId=1146029695717560320, journalId=1302319053441957962, issueId=1322631228000424183, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null, downloadFileDto=null}, startPage=7321, endPage=7333, ext={EN=ArticleExt(id=1322631261663908232, articleId=1322631260816658821, tenantId=1146029695717560320, journalId=1302319053441957962, language=EN, title=Research on quality control of soft material fine powder of Guizhi Fuling Capsules based on cascade fault-tolerant model, columnId=null, journalTitle=Chinese Traditional and Herbal Drugs, columnName=null, runingTitle=null, highlight=null, articleAbstract=Objective To address the potential uneven distribution of active ingredients indicated by uneven color within batches of soft material fine powder during the production process of Guizhi Fuling Capsules (GFC, 桂枝茯苓胶囊), a “cascade fault-tolerant model” was constructed based on color difference characteristics to explore a new low-cost approach for intermediate quality control. Methods A total of 177 samples of soft material fine powder from five batches were collected. The paeoniflorin content, determined by HPLC, was divided into low, medium, and high grades at a ratio of 2:6:2. Based on a traditional three-class model, a cascade fault-tolerant quality control model was constructed by comparing 126 algorithm combinations, which involved a primary algorithm for early warning and a secondary physical direction for correction. External validation was introduced to assess the generalization ability, and quality control experiments were conducted based on the model results. Finally, near-infrared spectroscopy (NIRS) technology was used for benchmarking against the upper limit of efficacy. Results Correlation analysis established paeoniflorin as the core grading indicator, with its content showing a significant positive correlation with red-green chroma a * (r = 0.64, P < 0.001). The analysis indicated that the accuracy of the benchmark model was only 70.34% (RSD was 12.70%), which was difficult to meet quality control requirements. However, the cascade fault-tolerant model achieved an overall industrial accuracy of 78.62%, with significantly enhanced robustness (RSD was 6.79%), and its external validation accuracy was 80.56%. Quality control experiments showed that the RSD of paeoniflorin content in mixed samples decreased from 7.74% in random mixing to 2.34%, indicating a significant increase in material uniformity, and a simultaneous reduction in total mixing process time and load. Benchmarking results with NIRS (accuracy of 90.68%, RSD was 6.22%) indicated that the colorimeter had a high cost-performance ratio and could be used for batch screening; NIRS was more accurate and suitable for high-precision grading. Conclusion The cascade fault-tolerant model constructed in this study achieved the synergistic empowerment of algorithm early warning and physical correction, effectively enhancing the discriminative robustness of low-cost optical sensors. This provides a new strategy for quality control of active ingredients in batches of intermediates during the production process of traditional Chinese medicine in the future., authors=ZHONG Xiaohui1,2 , TONG Feng1,3 , HOU Huarui1,3 , ZHANG Yongchao1,3 , WANG Tuanjie1,3 , ZHANG Chenfeng3,4 , XU Fangfang1,3 , LIU Hengxu1,3 , authorsList=ZHONG Xiaohui, TONG Feng, HOU Huarui, ZHANG Yongchao, WANG Tuanjie, ZHANG Chenfeng, XU Fangfang, LIU Hengxu, authorCompany=1 State Key Laboratory of Chinese Medicine Pharmaceutical Process Control and Intelligent Manufacture Technology, Jiangsu Kanion Pharmaceutical Co., Ltd.,/Nanjing University of Chinese Medicine, Nanjing 211112, China; 目的 针对桂枝茯苓胶囊（Guizhi Fuling Capsules，GFC）生产过程中软材细粉批内颜色不均可能预示的功效成分分布不均，构建基于色差特征的“级联容错模型”，探索一种低成本中间体质控新路径。方法 收集5批共177个软材细粉样本，以HPLC法测得的芍药苷含量按2∶6∶2比例划分为低、中、高等级。以传统三分类模型为基准，对比126种算法组合，构建一级算法预警协同二级物理判向的级联容错质控模型；引入外部验证评估泛化能力，并依据模型结果进行质控实验；最后利用近红外光谱（near-infrared spectroscopy，NIRS）技术进行效能上限对标。结果 相关性分析确立芍药苷为核心分级指标，其含量与红绿色度（a *）呈显著性正相关（r ＝0.64，P ＜0.001）。分析结果表明，基准模型准确率仅为70.34%（RSD为12.70%），难以满足质控要求；而级联容错模型工业总准确率为78.62%，稳健性显著增强（RSD为6.79%），其外部验证准确率为80.56%。质控实验显示，混合样品的芍药苷含量RSD，由随机混合的7.74%降至2.34%，物料均匀度显著增加，总混工序工时与载荷同步降低。与NIRS（准确率为90.68%，RSD为6.22%）的对标结果表明，色差仪性价比高，可用于批量初筛；NIRS更精准，适用于高精度分级。结论 构建的级联容错模型实现了算法预警与物理纠偏的协同赋能，有效提升了低成本光学传感器的判别稳健性，为未来中药生产过程中批内中间体功效成分的质控提供了新策略。, authors=钟小慧1,2 , 童枫1,3 , 候化蕊1,3 , 张永超1,3 , 王团结1,3 , 章晨峰3,4 , 徐芳芳1,3 , 刘恒旭1,3 , authorsList=钟小慧, 童枫, 候化蕊, 张永超, 王团结, 章晨峰, 徐芳芳, 刘恒旭, authorCompany=1 中药制药过程控制与智能制造技术全国重点实验室, 江苏康缘药业股份有限公司/南京中医药大学, 江苏 南京 211112;
中草药
|药剂与工艺
2026
, 57
(18) :
7321
-7333
基于级联容错模型的桂枝茯苓胶囊软材细粉质量控制研究
全屏
钟小慧1,2 , 童枫1,3 , 候化蕊1,3 , 张永超1,3 , 王团结1,3 , 章晨峰3,4 , 徐芳芳1,3 , 刘恒旭1,3
作者信息
1 中药制药过程控制与智能制造技术全国重点实验室, 江苏康缘药业股份有限公司/南京中医药大学, 江苏 南京 211112;
通讯作者:
徐芳芳
作者简介:
钟小慧: 钟小慧，女，硕士研究生，研究方向为药物制剂与产品研发。E-mail:15680541248@163.com
Research on quality control of soft material fine powder of Guizhi Fuling Capsules based on cascade fault-tolerant model
ZHONG Xiaohui1,2 , TONG Feng1,3 , HOU Huarui1,3 , ZHANG Yongchao1,3 , WANG Tuanjie1,3 , ZHANG Chenfeng3,4 , XU Fangfang1,3 , LIU Hengxu1,3
Affiliations
1 State Key Laboratory of Chinese Medicine Pharmaceutical Process Control and Intelligent Manufacture Technology, Jiangsu Kanion Pharmaceutical Co., Ltd.,/Nanjing University of Chinese Medicine, Nanjing 211112, China;
doi: 10.7501/j.issn.0253-2670.2026.18.020
文章导航
目的 针对桂枝茯苓胶囊（Guizhi Fuling Capsules，GFC）生产过程中软材细粉批内颜色不均可能预示的功效成分分布不均，构建基于色差特征的“级联容错模型”，探索一种低成本中间体质控新路径。方法 收集5批共177个软材细粉样本，以HPLC法测得的芍药苷含量按2∶6∶2比例划分为低、中、高等级。以传统三分类模型为基准，对比126种算法组合，构建一级算法预警协同二级物理判向的级联容错质控模型；引入外部验证评估泛化能力，并依据模型结果进行质控实验；最后利用近红外光谱（near-infrared spectroscopy，NIRS）技术进行效能上限对标。结果 相关性分析确立芍药苷为核心分级指标，其含量与红绿色度（a *）呈显著性正相关（r ＝0.64，P ＜0.001）。分析结果表明，基准模型准确率仅为70.34%（RSD为12.70%），难以满足质控要求；而级联容错模型工业总准确率为78.62%，稳健性显著增强（RSD为6.79%），其外部验证准确率为80.56%。质控实验显示，混合样品的芍药苷含量RSD，由随机混合的7.74%降至2.34%，物料均匀度显著增加，总混工序工时与载荷同步降低。与NIRS（准确率为90.68%，RSD为6.22%）的对标结果表明，色差仪性价比高，可用于批量初筛；NIRS更精准，适用于高精度分级。结论 构建的级联容错模型实现了算法预警与物理纠偏的协同赋能，有效提升了低成本光学传感器的判别稳健性，为未来中药生产过程中批内中间体功效成分的质控提供了新策略。
桂枝茯苓胶囊
/
软材细粉
/
色差特征
/
级联容错
/
中间体质控
/
近红外光谱
/
效能对标
/
工业应用
Objective To address the potential uneven distribution of active ingredients indicated by uneven color within batches of soft material fine powder during the production process of Guizhi Fuling Capsules (GFC, 桂枝茯苓胶囊), a “cascade fault-tolerant model” was constructed based on color difference characteristics to explore a new low-cost approach for intermediate quality control. Methods A total of 177 samples of soft material fine powder from five batches were collected. The paeoniflorin content, determined by HPLC, was divided into low, medium, and high grades at a ratio of 2:6:2. Based on a traditional three-class model, a cascade fault-tolerant quality control model was constructed by comparing 126 algorithm combinations, which involved a primary algorithm for early warning and a secondary physical direction for correction. External validation was introduced to assess the generalization ability, and quality control experiments were conducted based on the model results. Finally, near-infrared spectroscopy (NIRS) technology was used for benchmarking against the upper limit of efficacy. Results Correlation analysis established paeoniflorin as the core grading indicator, with its content showing a significant positive correlation with red-green chroma a * (r = 0.64, P < 0.001). The analysis indicated that the accuracy of the benchmark model was only 70.34% (RSD was 12.70%), which was difficult to meet quality control requirements. However, the cascade fault-tolerant model achieved an overall industrial accuracy of 78.62%, with significantly enhanced robustness (RSD was 6.79%), and its external validation accuracy was 80.56%. Quality control experiments showed that the RSD of paeoniflorin content in mixed samples decreased from 7.74% in random mixing to 2.34%, indicating a significant increase in material uniformity, and a simultaneous reduction in total mixing process time and load. Benchmarking results with NIRS (accuracy of 90.68%, RSD was 6.22%) indicated that the colorimeter had a high cost-performance ratio and could be used for batch screening; NIRS was more accurate and suitable for high-precision grading. Conclusion The cascade fault-tolerant model constructed in this study achieved the synergistic empowerment of algorithm early warning and physical correction, effectively enhancing the discriminative robustness of low-cost optical sensors. This provides a new strategy for quality control of active ingredients in batches of intermediates during the production process of traditional Chinese medicine in the future.
Guizhi Fuling Capsules
/
soft material fine powder
/
color difference feature
/
cascaded fault tolerance
/
intermediate product quality control
/
near-infrared spectroscopy
/
performance benchmarking
/
industrial application
钟小慧, 童枫, 候化蕊, 张永超, 王团结, 章晨峰, 徐芳芳, 刘恒旭.
基于级联容错模型的桂枝茯苓胶囊软材细粉质量控制研究.
中草药,
2026
, 57
(18)
: 7321
-7333
.
DOI: 10.7501/j.issn.0253-2670.2026.18.020
ZHONG Xiaohui, TONG Feng, HOU Huarui, ZHANG Yongchao, WANG Tuanjie, ZHANG Chenfeng, XU Fangfang, LIU Hengxu.
Research on quality control of soft material fine powder of Guizhi Fuling Capsules based on cascade fault-tolerant model[J].
Chinese Traditional and Herbal Drugs ,
2026
, 57
(18)
: 7321
-7333
.
DOI: 10.7501/j.issn.0253-2670.2026.18.020
参考文献
引证文献
国家药品监督管理局. 国家药监局关于发布中药生产监督管理专门规定的公告（2025年第79号）[EB/OL]. (2025-09-08) [2026-09-10]. https://www.nmpa.gov.cn/ xxgk/ggtg/ypggtg/ypqtggtg/20250908094819150.html.
2026年第57卷第18期
PDF下载
32
4
引用本文
BibTeX
文章信息
doi: 10.7501/j.issn.0253-2670.2026.18.020
接收时间：2026-03-31
首发时间：2026-10-29
https://castjournals.cast.org.cn/joweb/zcy/CN/10.7501/j.issn.0253-2670.2026.18.020
引用本文
BibTeX
2种不同金属材料的力学参数
科 属数 种数 占总种数比例 属 种数 占总种数比例 鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78 小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39 多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39 红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87 小菇属 Mycena 11 5.26 光柄菇属 Pluteus 5 2.39 红菇属 Russula 17 8.13 栓菌属 Trametes 5 2.39
关闭全屏
BibTeX
EndNote
RefWorks
TxT
PDF下载
全屏
32
4
关闭全屏
北京市海淀区学院南路86号
010-62662699（运营及技术支持）
qkjq@cast.org.cn
