Page 53 - 《中国药房》2025年19期
P. 53
Box-Behnken响应面法对比GA-BP神经网络优化知母盐炙工艺
Δ
1 #
2
2
2
3
潘罗星 ,赵一曼 ,袁 慧 ,李泽华 ,薛东升 ,赵 清 (1.河北大学中医学院河北省中医药管理局中药资源产
1*
业化过程协同创新重点研究室,河北 保定 071000;2.河北百合健康药业有限公司,河北 保定 071299;3.保
定市市场监督管理局,河北 保定 071023)
中图分类号 R943.1 文献标志码 A 文章编号 1001-0408(2025)19-2399-05
DOI 10.6039/j.issn.1001-0408.2025.19.07
摘 要 目的 优化知母的盐炙工艺。方法 以闷润时间、炒制温度及炒制时间为考察因素,以芒果苷、新芒果苷、异芒果苷、知母
皂苷BⅡ、知母皂苷AⅢ、知母皂苷BⅢ、总黄酮、总皂苷含量为考察指标,采用Box-Behnken响应面法设计实验,利用熵权法确定
各指标权重,并计算综合评分,得Box-Behnken响应面法优化的知母盐炙工艺。以17组Box-Behnken响应面法结果为基础,以闷
润时间、炒制温度、炒制时间为输入层,以综合评分为输出层,采用遗传算法(GA)-反向传播(BP)神经网络优化知母盐炙工艺。验
证并比较两种方法所得盐炙工艺参数,以确定知母的最佳盐炙工艺。结果 经Box-Behnken响应面法优化的知母盐炙工艺条件为
闷润时间 23 min、炒制温度 160 ℃、炒制时间 12 min,综合评分为 63.370 2 分;GA-BP 神经网络优化后的工艺条件为闷润时间 24
min、炒制温度163 ℃、炒制时间12 min,综合评分为65.163 8分。GA-BP神经网络优化结果优于Box-Behnken响应面法所得结果。
结论 本研究成功优化了知母的盐炙工艺,具体为每 50 g 饮片加入 0.1 g/mL 盐水 15 mL,闷润 24 min 后,在 163 ℃条件下炒制 12
min。
关键词 知母;盐炙;工艺优化;Box-Behnken响应面法;反向传播神经网络;遗传算法
Optimization of salt-processing technology for Anemarrhena asphodeloides by Box-Behnken response
surface methodology versus GA-BP neural network
PAN Luoxing ,ZHAO Yiman ,YUAN Hui ,LI Zehua ,XUE Dongsheng ,ZHAO Qing(1. Key Laboratory of
1
3
2
1
2
2
Collaborative Innovation in the Industrialization Process of Traditional Chinese Medicine Resources of Hebei
Provincial Administration of Traditional Chinese Medicine, College of Traditional Chinese Medicine, Hebei
University, Hebei Baoding 071000, China;2. Hebei Baihe Health Pharmaceutical Co., Ltd., Hebei Baoding
071299, China;3. Baoding Market Supervision Administration, Hebei Baoding 071023, China)
ABSTRACT OBJECTIVE To optimize the salt-processing technology for Anemarrhena asphodeloides. METHODS Taking
soaking time, stir-frying temperature, and stir-frying time as factors, Box-Behnken response surface methodology was employed to
optimize the salt-processing technology of A. asphodeloides using the contents of mangiferin, neomangiferin, isomangiferin,
timosaponin BⅡ, timosaponin AⅢ, timosaponin BⅢ, total flavonoids, and total saponins as evaluation indicators. The entropy
weight method was applied to determine the weight of each indicator and calculate the comprehensive score. Based on the 17 sets of
Box-Behnken response surface methodology results, a genetic algorithm (GA)-back propagation (BP) neural network was used to
further optimize the salt-processing technology, with soaking time, stir-frying temperature, and stir-frying time as input layers and
the comprehensive score as the output layer. The salt-processing parameters obtained from the two methods were validated and
compared to determine the optimal salt-processing technology for A. asphodeloides. RESULTS The optimal salt-processing
conditions obtained via the Box-Behnken response surface methodology were as follows: soaking time of 23 min, stir-frying
temperature of 160 ℃, and stir-frying time of 12 min, yielding a comprehensive score of 63.370 2. The GA-BP neural network
optimization resulted in the following conditions: soaking time of 24 min, stir-frying temperature of 163 ℃, and stir-frying time of
12 min, with a comprehensive score of 65.163 8. The GA-BP neural network optimization outperformed the results obtained by
Box-Behnken response surface methodology. CONCLUSIONS This study successfully optimized the salt-processing technology for
A. asphodeloides. Specifically, the technology involves adding 15 mL of 0.1 g/mL saline solution to 50 g of the herbal slices,
allowing them to moisten for 24 minutes, and then stir-frying
Δ 基金项目 河北省重点研发计划项目-中医药创新专项(No.
at 163 ℃ for 12 minutes.
22372504D) KEYWORDS
*第一作者 硕士研究生。研究方向:中药饮片炮制工艺与炮制机 Anemarrhena asphodeloides; salt-processing
理。E-mail:lxtd1208@163.com technology; technology optimization; Box-Behnken response
# 通信作者 副教授,硕士生导师,博士。研究方向:中药饮片炮制 surface methodology; back propagation neural network;
工艺与炮制机理。E-mail:wshxr2003@163.com genetic algorithm
中国药房 2025年第36卷第19期 China Pharmacy 2025 Vol. 36 No. 19 · 2399 ·

