基于遗传神经网络的郫县豆瓣有机酸提取工艺研究

针对郫县豆瓣有机酸最佳提取工艺问题,提出了一种高精度遗传神经网络优化方法,解决回归分析方法拟合度精度和准确性不高的现象。以郫县豆瓣有机酸提取量为性能指标,用中心组合试验数据训练遗传神经网络,当神经网络训练误差达到1.1862×10^-11时,神经网络算法输出数据与试验数据几乎完全拟合,平均相对误差为0%,而回归模型为0.822%,对比分析得出遗传神经网络算法的拟合精度和拟合优度均高于回归模型。用训练好的遗传神经网络算法对郫县豆瓣有机酸提取工艺条件进行优化,得到最佳提取工艺条件为:乙醇体积分数68.50%,料液比1∶20.44(g/mL),超声时间35.28min。此时有机酸提取量达到最大值15.19mg/g。该方法可为郫县豆瓣有机酸提取工艺提供试验参考,为制定更加完善的产品质量标准提供了数据支撑。

Study on Extraction Process of Organic Acid from Pixian Broad Bean Paste Based on Genetic Neural Network

According to the problem of the optimal extraction process of organic acid in Pixian broad bean paste,a high-precision genetic neural network optimization method is proposed to solve the phenomenon that the fitting accuracy and accuracy of regression analysis method are not high.The extraction amount of organic acid in Pixian broad bean paste as the performance index,the genetic neural network is trained by the central combination test data.When the training error of neural network reaches 1.1862×10^-11,the output data of the neural network algorithm is almost completely fitted with that of the experiment,and the average relative error is 0%,while the average relative error of the regression model is 0.822%.The fitting accuracy and goodness of the genetic neural network algorithm are better than those of the regression model by comparative analysis.When the trained genetic neural network algorithm is used to optimize the extraction process parameters of the organic acid in Pixian broad bean paste,the optimal extraction process parameters are as follows:ethanol volume fraction is 68.50%,the solid-liquid ratio is 1∶20.44(g/mL),and ultrasonic time is 35.28 min,the extraction amount of organic acid reaches a maximum of 15.19 mg/g.This method can provide a test reference for the extraction process of organic acid from Pixian broad bean paste,and provide data support for formulating more perfect product quality standards.