摘要
选择三层结构的BP神经网络,建立镁合金微弧氧化膜厚的映射模型。基于采用带放电回路的脉冲电源进行的镁合金微弧氧化的实验结果,讨论了微弧氧化的几个主要参数对膜层生长的影响关系,从而确定了神经网络的输入参数。通过工艺试验构造了训练与测试样本集。训练得到的BP神经网络的测试结果表明,利用神经网络的方法来实现对膜厚的动态监测是可行的。
The mapping model of film thickness of micro-arc oxidation（MAO） on magnesium alloys is built on BP neural network with three layer structure.Based on experimental results of magnesium alloy MAO under the pulse power supply with discharge loop,the affecting relations between film thickness and several main parameters of MAO were put forward,thus causing the input parameters of neural network to be determined.Then the training and testing samples were constructed through technical experiment.Finally,the testing results of BP neural network which was got through training proved that the method of dynamic monitoring of film thickness of MAO based on neural network was feasible.
出处
《铸造技术》
CAS
北大核心
2011年第6期845-848,共4页
Foundry Technology
基金
甘肃省科技重大专项资助项目(0702GKDA024)
关键词
镁合金
微弧氧化
膜层厚度
神经网络
Magnesium alloys
Micro-arc oxidation
Film thickness
Neural network
