关键词: 胡椒, 荜茇, 熵权法, Box-Behnken响应面法, 谱效关系, 灰色关联度, BP神经网络

Abstract: Objective: To optimize the extraction process of Hujiao-Biba by using Box-Behnken response surface method combined
with BP neural network based on spectrum-effect relationship and entropy weight method. Methods: Effects of solid-liquid ratio,
extraction times and ultrasonic time on the ultrasonic extraction process of Hujiao-Biba were investigated by using Box-Behnken
response surface method, and the yield of dry extract and DPPH· clearance rate were calculated. HPLC fingerprint of Hujiao-Biba was
established and grey correlation method was used to determine the spectrum-effect relationship between fingerprint and DPPH·
clearance rate, and the correlation degree was calculated. The correlation degree was used to correct the peak area of piperine and the
total peak area of the fingerprint, and the corrected peak area was obtained. The weight of each evaluation index was assigned by
entropy weight method, and the comprehensive evaluation index was calculated. A BP neural network model was established to
optimize the experimental results and the optimal extraction process for Hujiao-Biba was ultimately obtained. Results: The optimal
extraction process for Hujiao-Biba obtained through Box-Behnken response surface method and BP neural network model prediction
was as follows: the solid-liquid ratio was 1 : 30 （g/mL）, the extraction times was 4 times and the ultrasonic time was 40 min. Conclusion: Box-Behnken response surface method combined with BP neural network model based on spectrum-effect relationship
and entropy weight method can be used to optimize the extraction process of Hujiao-Biba, laying a foundation for the preparation of
their extract and the study of related dosage forms.

Key words: Hujiao, Biba, entropy weight method, Box-Behnken response surface method, spectrum-effect relationship, gray correlation;
BP neural network