中文摘要：

为研究甘草中主要生理活性物质甘草酸的超声提取过程，在正交试验的基础上，建立了混合神经网络模型，包括基本原理模型和由神经网络组成的模型参数估值器．以正交试验数据对混合神经网络模型进行了训练，获得了较好的过程模拟结果．以此为基础，得到了超声提取甘草酸的优化工艺条件，即液固比为10mL／g、提取温度25℃、浸泡时间150min、超声时间60min．该优化工艺条件操作简便、能耗低、提取率高，合理可行．

英文摘要：

The supersonic assisted extracting technology of glycyrrhizic acid from liquorice was studied. On the basis of orthogonal experiment,a hybrid neural networks （HNN） model was proposed to predict the glycyrrhizic acid extraction rates. The model consisted of a first principal partial model and a neural network parameter estimator. The model was trained with the orthogonal experimental data and had a good performance of process simulation. By using this new method, the optimum extraction conditions were obtained as follows.ratio of liquid-solid 10 mL/g,extracting temperature 25℃ ,marinating time 150 min and supersonic time 60 min. With optimizing extraction technology by HNN,the experimental information could be dig out completely. Applying this preferable extracting technology, the costs of production is reduced, the energy consuming is decreased and the glycyrrhizic acid content is increased. The resuits showed that this new method is reasonable and practicable.