摘要
研究并应用径向基函数神经网络(RBF-network)对稀土卤化物标准生成焓进行定量建模,相关系数r达到0.9985,样本误差较小,显现了神经网络方法在该领域的优势.交互预测结果证明模型具有好的稳定性和泛化性能,对7个未知样本所作的预测有一定参考价值.
A RBF- Network was applied to study the relationships between the Standard Formative Enthalpy of 57 Halide of rare earth and the structural parameters. The results obtained show a good relationship between the calculated and experimental ΔfHmΘ data with a fitting correlation coefficient 0.9985, which is better than that by linearly method, and exhibit advantage of the neural network method in this field. The result of mutual predict prove that the model has good stability and generalization ability, The predicted values of 7 unknown samples are to a certian extent valuable.
出处
《怀化学院学报》
2005年第2期46-48,共3页
Journal of Huaihua University
基金
湖南省高校科研资助项目(No.01C035).
