基于机器学习的马尾松松花粉提取工艺研究

为提高马尾松松花粉提取率，对采用正交实验法和机器学习法提取马尾松松花粉的效果进行了对比研究.研究显示：以马尾松松花粉浸膏为指标时，机器学习方法获得的马尾松松花粉浸膏的优化验证值为21.06%,正交试验的验证值为20.98%;以松花粉多糖为指标时，机器学习方法获得的多糖提取率的优化验证值为2.74%,正交试验的验证值为2.61%;以松花粉综合评分为指标时，机器学习方法获得的综合评分优化验证值为92.01%,正交试验的验证值为89.78%.该结果表明，采用机器学习法提取马尾松松花粉浸膏和多糖的效果显著优于正交实验法，同时该方法可为同类提取工艺提供参考.

Research on optimization of Pinus massoniana Lamb extraction process based on machine learning

To improve the extraction rate of Pinus massoniana Lamb, the orthogonal experiment method and machine learning method were adopted and compared.According to the results, when the Pinus massoniana Lamb extract was taken as the index, the optimized validation value of the Pinus massoniana Lamb extract obtained by the machine learning method reached 21.06%, and the validation value of the orthogonal test reached 20.98%.In addition, taking polysaccharide of the Pinus massoniana Lamb as an index, the optimized validation value of the extraction rate of polysaccharide obtained through the machine learning method was 2.74%, and the validation value of orthogonal test reached 2.61%.Furthermore, taking the comprehensive score of the Pinus massoniana Lamb as the index, the optimized verification value of the comprehensive score obtained by the machine learning method was 92.01%, and the validation value of the orthogonal test reached 89.78%, which shows that the machine learning optimization results are better than the orthogonal experiment results in extraction rate of extract and polysaccharides.At the same time, this method can provide reference for the extraction of similar processes.