基于RBF神经网络RE-Ni-Cu合金铸铁静态腐蚀性能预测

Static corrosion prediction of RE-Ni-Cu alloy cast iron based on RBF neural network

通过静态腐蚀试验获取35组样本数据,利用MATLAB软件的工具箱函数建立RBF神经网络预测模型,并对RENi-Cu合金铸铁的静态腐蚀深度和耐蚀性进行预测研究。结果表明:RBF神经网络预测RE-Ni-Cu合金铸铁在浓碱液中的静态腐蚀性能可行且有效,能较好地反映主要合金成分、腐蚀时间、碱液温度与静态腐蚀深度之间的非线性映射关系;当RBF网络的扩展系数为0.5,静态腐蚀深度的网络预测值与实测值之间的相对误差最小,且耐蚀等级和耐蚀评价的准确率均达到100%。

The 35 groups sample data were measured by the static corrosion test. The RBF neural network prediction model was established by the toolbox function of Matlab,and the corrosion depth and corrosion resistance of RE-Ni-Cu alloy cast iron were predicted. The results show that RBF neural network can effectively predict the static corrosive nature of RE-Ni-Cu alloy cast iron in concentrated alkaline solution,and it reflects the non-linear relationship between main components of alloy cast iron and corrosion time and alkaline solution temperature and static corrosion depth very well. When the spread of RBF neural network is0.5,the error between predicted values of network and measured values of static corrosion depth is least,and the accuracy rate of corrosion resistanc grade and corrosion resistance appraisal reaches 100%.