论文部分内容阅读
设计了一个基于实数编码的改进进化算法优化神经网络的连接权和网络结构.该算法可以根据种群停止进化代数自适应调节变异率、根据个体适应度调节变异量.加工实验表明采用进化神经网络可以较准确预测出电火花铣削加工工具损耗,所提出的进化算法是有效的,预测结果较标准BP神经网络高.该预测模型为电火花铣削加工工具在线自动补偿打下基础.
This paper designs an improved evolutionary algorithm based on real number coding to optimize the connection weights and network structure of neural networks.The algorithm can adaptively adjust the mutation rate according to the population evolution evolution algebra and adjust the variation according to individual fitness.Experimental results show that using evolutionary neural network The prediction of EDM is more effective than the standard BP neural network, which predicts the loss of EDM tool. The prediction model is the foundation for EDM tool online automatic compensation.