神经网络的遗传设计及其在甲醛生产建模及优化中的应用

采用遗传算法的机制设计了一种同时对前向网络的连接权值及拓扑结构进行搜索的方法，并采用以隐结点为基本基因型的方式，同时在算法中加入了局部搜索的方法．该算法已被应用于甲醛生产过程的建模及优化，达到了良好的效果．

Genetic Evoloved Neural Network and Its Application in Formaldehyde Process Modeling and Optimization

The artificial neural network has been widely apllied for its ability of learning from examples. Up to now, most of the literature deals only with the learning of connection weights in a network, but few has been done for the structure of a network. A GA-based training algorithm with a neuron-wise strategy which is capable of finding optimal structure and connection weights of network is proposed in this paper; enhancement of GA by adding local search ability is also proposed. Application of this algorithm to the modeling and optimization of an industrial formaldehyde production process shows the encouraging characteristics of this algorithm.