咖啡因乙醇溶解度的神经网络模型建立的研究

The research on the establishment of artificial nevual network model based on the solubility of caffeine in ethanol

根据实验室提供咖啡因热力学实验数据,经过筛选选择了较好的人工神经网络模型2-2-1BP神经网络模型,利用训练后的2-2-1BP人工神经网络模型对数据进行了内插和外推的预测,在乙醇溶解度数据的内插的误差均为4％以内,外推的误差在3％以内,效果十分令人满意。

Based on the thermodynamic data of caffeine provided by laboratory, a 2 - 2- 1 backpropagation (BP)artificial neural network (ANN) model is selected from many other ANN models. The prediction of interposition and extrapolation of the data has been taken with the trained 2 - 2 -1BP ANN model. The outcome is very satisfactory.