GA-BP神经网络的钛合金微弧氧化膜层厚度预测模型的建立.pdfVIP

向;电镜萄园除S审 暗协第7 期 Vol.34 No.7 ,_, E幅民幅画ng ‘同，幅画曲唱 GA-BP 神经网络的铁合金微弧氧化膜层厚度预测模型的建立 牛宗伟*，李明哲 (山东理工大学机械工程学院，山东淄博 255000) 摘要:利用正文试验法获得的 TC4 钦合金微弧氧化实验数据建立了基于 4-11-1 (即 4 个输入神经元， II 个隐含层节点， 1 个输出 神经元)结构的 BP 神经网络预测膜层厚度的模型，并引入遗传算法(GA) 对其权值和l到值进行优化。以微弧氧化工艺参数中的电流 密度、脉冲频率、占空比和氧化时间作为网络的输入向量，氧化膜层厚度作为网络的输出向量，对比和分析了 BP 与 GA-BP 模型 的预测结果。与 BP 网络模型相比， GA-BP 网络模型稳定性能较好，并能高精度预测膜层的厚度， GA-BP 网络模型预测值的平均 误差为 0.015 ，最大误差仅为 0.036 ，而 BP 模型预测结果的平均误差为 0.064 ，最大误差为 0.0990 关键i司z 钦合金;微弧氧化;膜辱;预测;神经网络;遗传算法 中图分类号: TG174; TPI83 文献标志码:A 文章编号: 1004 - 227X (2015) 07 一 0381 -05 Establishment of GA-BP neural network model for prediction of the thickness of micro-arc oxidation coating on titanium alloy // NIU Zong-wei 气 LI Ming-zhe Abstract: A modeI based on BP (back propagation) neuraI network with 4- 11- 1 (i.e. 4 input neurons , 11 hidden Iayer nodes , and 1 output neuron) structure for prediction of the thickness of micro-arc oxidation coating on TC4 titanium alloy was established using the data obtained by orthogonaI test. The weights and thresholds were optimized by genetic aIgorithm (GA). The prediction results ofthe BP and GA-BP modeIs were analyzed and compared using current density , frequency , duty cycle , and oxidation time as input vectors and the thickness of oxidation coating as output vector. Compared to BP model , the GA-BP model possesses better stability and predicts the coating thickness with higher precision. The average and maximum prediction e町ors are 0.015 and 0.036 respectively for GA-BP model , and 0.064 and 0.099 respectively for BP mode l. Keywords: titanium alloy; micro-arc oxidation; film thickness; prediction; neural network; genetic algorithm First-authors address: School