铁水硅预报神经网络专家系统的遗传优化生成

利用遗传算法所固有的全局搜索性能，进化多层前馈神经网络连接权，结合ＢＰ学习算法和符号主义专家系统，建立了高炉铁水硅含量预报遗传算法神经网络专家系统。该系统解决了神经网络局部的收敛、学习时间过长的问题。实际应用表明该系统可以提高硅预报命中率，并能清楚解释推理过程，提供高炉操作指导。

OPTIMIZING GENERATION OF AN EXPERT SYSTEM BASED ON NEURAL NETWORK BY GENETIC ALGORITHM TO PREDICT THE SILICON CONTENT IN HOT METAL

An expert system based on a genetic algorithms with fast global convergence, which is able to evolve connection weights of a multi layer feed forward neural network and combine BP learning algorithms, is introduced in this paper The system has been applied in prediction of the silicon content in hot metal The problem from local convergence and much time in learning was resolved The practical result shows that the hitting rate of predicting the silicon content has been increased, the reasoning process can clearly be explained and the system can be used as operation guide on BF