摘要
用前向神经网络 ,对纯物质的蒸气压和汽化热与温度的函数关系进行预测。通过适当变量变换 ,在相同网络单元数情况下 ,大大提高预测精度。对 387种物质的预测结果表明 :在熔点到临界点的温度范围内 ,蒸气压的平均预测误差为 0 .0 84% ,汽化热的平均预测误差为0 .0 1 8%。
A feedforward neural network was used to predict functional relationship of temperature with steam pressure and heat of vaporization of the pure material.By varying suitably variables,a high degree of accuracy was received at the network of the same units. Predicted results based on 387 cases showed that the average estimated error of stream pressure and heat of vaporization were respectively 0 084% and 0 018% in the temperature ranging from the melting point to critical point.
出处
《广西科学院学报》
2001年第1期14-17,共4页
Journal of Guangxi Academy of Sciences