Optimization of Extraction Process of Aqueous Extract of Rehmannia glutinosa Based on Entropy Weight Method in Cooperation with Analytic Hierarchy Process and Back Propagation Neural Network with Multiple Indicators

Objective: To optimize the ultrasonic extraction process of aqueous extract of Rehmannia glutinosa based on entropy weight method (EWM) together with analytic hierarchy process (AHP) and back propagation neural network (BPNN). Methods: Taking ultrasonic temperature, ultrasonic time, solid-liquid ratio as orthogonal test factors, catalpa, rehmannioside D, ajugol, rehmannioside polysaccharide content, and the yield of water-soluble extract as indexes, EWM-AHP method was used to determine the weight of each index, to obtain the comprehensive score of multiple indexes. Then, the experimental process conditions were used as the input and the synthesis score was used as the output to build and train a BPNN model to find the best process for the extraction process of Rehmannia glutinosa. Results: The preferred process of BPNN was extraction at 60 ℃ with 33 times of water for 70 min. The mean values of catalpa, rehmannioside D, ajugol, rehmannioside polysaccharide content, and the yield of water-soluble extract were 17.04, 3.75, 10.57, 24.86 mg/g and 0.82 g/g, respectively, with a combined score of 97.74; The best process of the orthogonal test was extraction at 50 ℃ with 25 times of water for 1 h. The mean values of catalpa, rehmannioside D, ajugol, rehmannioside polysaccharide content, and the yield of water-soluble extract were 12.72, 2.58, 8.20, 25.02 mg/g and 0.80 g/g, respectively, with a combined score of 97.26. Conclusion: This study determined that the BPNN preferred process was optimal and provided a reference for the application of the EWM-AHP method combined with BPNN in the extraction process.