CAJ | 학술논문

为径向基神经网络确定更为优化的初始中心，增强径向基网络的性能。通过采用改进的模拟退火蚁群算法作为径向基神经网络径向基层的训练法，将改进的径向基神经网络模型应用于地层高程的面插值和矿体品位的空间体插值，并与普通克里金法进行交叉验证，优化效果明显，然后利用VC＋＋与OpenGL开发环境开发出矿体可视化系统，结果在结合矿体实际数据进行实例应用的过程中，实效性明显。
위경향기신경망락학정경위우화적초시중심，증강경향기망락적성능。통과채용개진적모의퇴화의군산법작위경향기신경망락경향기층적훈련법，장개진적경향기신경망락모형응용우지층고정적면삽치화광체품위적공간체삽치，병여보통극리금법진행교차험증，우화효과명현，연후이용VC＋＋여OpenGL개발배경개발출광체가시화계통，결과재결합광체실제수거진행실례응용적과정중，실효성명현。
The improved simulated annealing ant colony algorithm is used as the radial basic training method of RBF neural network.Its more optimization determines the initial center for radial basis neural network and enhance the performance of the radial basis network.The improved RBF neural network model has been applied to the stratigraphic vertical surface interpolation and orebody spatial interpolation,and verified crossly with the ordinary Kriging method,and the optimized effect is obvious.Then,the ore body visualization system developed by VC++ and OpenGL development environment software is used.The result shows that in the application of the example combined with the actual data of the orebody,the effectiveness is very obvious.