众多学者研究的两类形态学联想记忆网络的存储能力、抗腐蚀/膨胀噪声的能力等性质几乎都相同.但是文中研究发现两类网络对训练模式摄动的鲁棒性差异很大.一类对训练模式摄动拥有好的鲁棒性,而另一类则较差.该研究结论能为形态学联想记忆网络的学习算法选择和训练模式采集设备的精度要求提供指导,对前期训练模式的获取过程提供警示. Many kinds of morphological associative memory networks studied by many scholars have almost the same storage capacity and ability of anti-corrosion / expansion noise, but the research shows that the robustness of the two types of networks to the perturbation of training mode is quite different. Which has good robustness to perturbations in training mode and poor in the other class.The results of this study can guide the selection of training algorithms and the precision requirements of training mode acquisition devices in morphological associative memory networks, Get the process to provide a warning.