基于神经网络的金盏花粗多糖提取工艺优化、理化性质及抗氧化活性

Optimization of Extraction Process, Physicochemical Properties, and Antioxidant Activity of Calendula officinalis Crude Polysaccharides Based on Neural Networks

摘要:

为实现金盏花多糖的高效开发与利用，本研究通过单因素实验获取数据集，以多糖得率为评价指标，分别采用人工神经网络（Artificial Neural Networks，ANN）和液态神经网络（Liquid Neural Networks，LNN）建立预测模型，优化金盏花粗多糖的热水浸提工艺，并对最优工艺下所得多糖进行初步的理化性质与体外抗氧化活性分析。结果表明，两种神经网络模型均成功建模，LNN模型的泛化能力优于ANN，确定的最优提取条件为：液料比44:1（mL?g-1）、提取温度94℃、提取时间2.4 h。该条件下，多糖得率达3.37%，为单因素最优条件下的1.19倍。金盏花粗多糖总糖含量为42.08%，糖醛酸含量为29.30%，蛋白含量为8.32%，具有典型的多糖红外特征吸收峰，微观形貌呈多孔薄片状。体外抗氧化实验表明，在1 mg?mL-1浓度下，多糖对2,2-联苯基-1-苦基肼基（2,2-Diphenyl-1-picrylhydrazyl，DPPH）和2,2′-联氮双（3-乙基苯并噻唑啉-6-磺酸）二铵盐（2,2′-Azino-bis(3-ethylbenzothiazoline-6-sulfonic Acid) Diammonium Salt，ABTS）清除率分别为62.97%和84.65%，其IC50值均为0.5 mg?mL-1，铁离子还原能力为1.02 mmol?L-1。本研究为金盏花多糖在功能食品领域的进一步开发与应用提供了理论基础和方法参考。

Abstract:

In order to achieve the efficient development and utilization of Calendula officinalis polysaccharides, a dataset was obtained through single-factor experiments, with polysaccharide yield used as the evaluation index. Predictive models were established using Artificial Neural Networks (ANN) and Liquid Neural Networks (LNN) to optimize the hot water extraction process of Calendula officinalis crude polysaccharides. Under the optimal extraction conditions, the physicochemical properties and in vitro antioxidant activities of the obtained polysaccharides were further investigated. The results demonstrated that both neural network models were successfully established, with the LNN model exhibiting superior generalization performance compared with the ANN model. The optimal extraction conditions were determined as a liquid-to-solid ratio of 44:1 (mL?g-1), an extraction temperature of 94°C, and an extraction time of 2.4 hours. Under these conditions, the polysaccharide yield reached 3.37%, which was 1.19-fold higher than that obtained under the optimal single-factor conditions. The Calendula officinalis crude polysaccharides exhibited a total sugar content of 42.08%, a uronic acid content of 29.30%, and a protein content of 8.32%. Fourier transform infrared spectroscopy revealed characteristic absorption bands typical of polysaccharides, while scanning electron microscopy showed a porous, sheet-like microstructure. In vitro antioxidant assays demonstrated that, at a concentration of 1 mg?mL-1, the polysaccharides exhibited scavenging rates of 62.97% and 84.65% against 2,2-Diphenyl-1-picrylhydrazyl (DPPH) and 2,2′-Azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt (ABTS), respectively, with IC50 values of 0.5 mg?mL-1 for both, and a ferric reducing antioxidant power of 1.02 mmol?L-1. This study provides a theoretical basis and methodological reference for the further development and application of Calendula officinalis polysaccharides in the field of functional foods.