论文部分内容阅读
以化学键参数作为人工神经网络的输入，实测相图数据为输出，将采用误差反向传播算法训练好的神经网络用于对未知相图作计算机预报，将数据库与知识库相结合，设计和开发出了一个检索和预报二元及部分三元熔盐系相图特征的专家系统。该数据库包括各类已知熔盐相图特征的实验数据及熔盐系各种元素的化学键参数，而知识库为训练好的人工神经网络，通过人机对话形式提供相图特征的各种信息。给出了３个实际应用例子，对预报结果进行的实验验证表明该专家系统对未知相图特征的预报是可靠的。
Taking the chemical bond parameters as the input of artificial neural network and the measured phase diagram data as the output, the neural network trained by error back propagation algorithm is used to predict the unknown phase diagram by computer. The database and knowledge base are combined to design and develop An expert system for the retrieval and prediction of phase diagrams of binary and partial ternary molten salt systems was developed. The database includes the experimental data of all kinds of known molten salt phase diagrams and the chemical bond parameters of various elements of the molten salt system. The knowledge base is a trained artificial neural network that provides various information of the phase diagram features through man-machine dialogue . Three practical examples are given. The experimental verification of the forecast results shows that the expert system is reliable in forecasting the unknown phase diagram features.