摘要
采用超声耦合亚临界水提取香菇多糖,用人工神经网络技术建立提取数学模型,以提取温度、提取时间、提取压力、料液比、超声功率作为网络输入,香菇多糖得率作为输出,首先对亚临界水提取和超声耦合亚临界水提取单因素条件下的香菇多糖得率进行模拟预测,然后利用Design-expert软件对影响因素进行优化实验,分别用响应曲面法和人工神经网络进行多糖得率的模拟预测对比,并对最优化条件进行实验验证。结果表明,人工神经网络拟合值与实验值能很好的吻合,其拟合效果在一定程度上优于响应曲面。
The artificial neural network（ ANN） technology was used to simulate the mathematical model of lentinan polysaccharides（ LP）,which was extracted by the interpolated ultrasonically-coupled subcritical water extraction equipment. The extraction temperature,extraction time,extraction pressure,solid to liquid ratio and ultrasonic power were as input of the network,LP yield was as the output of the network. Firstly,using the network predicted the yield of LP on the condition of single factor by subcritical water extraction（SWE） and ultrasonically-coupled subcritical water extraction（USWE）. Then,Design-expert was used to optimize the experiment on the influence factor,the response surface method（ RSM） and artificial neural network（ ANN） was used to predict the yield of LP,respectly. The experiment was conducted on the best condition to verify. The result indicated the value predicted by ANN matched well with the experiment value,the simulation effect was better than RSM in some extent.
出处
《应用化工》
CAS
CSCD
北大核心
2015年第7期1372-1376,共5页
Applied Chemical Industry
基金
广东省省部产学研结合项目(2013B090600034)
关键词
神经网络
超声耦合亚临界水萃取
香菇多糖
响应曲面法
artificial neutral network
ultrasonically-coupled subcritical water extraction
lentinus polysaccharides
response surface methodology
