摘要
采用微弧氧化技术制备钛合金微弧氧化膜。通过正交试验得到了不同电解液组成下的膜层厚度,借助Matlab软件建立4-11-1的BP神经网络模型,利用遗传算法对网络的权值与阈值进行优化。优化后的网络能够较好地反映电解液参数与膜层厚度间的内在规律。与BP神经网络模型相比,GA-BP神经网络模型的预测精度更高。
Ceramic coatings were prepared on titanium alloy by micro-arc oxidation technology. The thickness of the coatings obtained in different electrolyte systems was determined through orthogonal experiment, and the network weights and thresholds were optimized based on the 4-11-1 BP neural network established by Matlab. Results showed that the optimized neural network can reflect the inherent laws between electrolyte parameters and coatings thickness. And furthermore, compared to BP model, the GA-BP model has higher precision.
出处
《电镀与环保》
CAS
CSCD
北大核心
2015年第3期42-44,共3页
Electroplating & Pollution Control
基金
国家自然科学基金(No.51005140)
山东省自然科学基金(No.ZR2010EQ037)
山东理工大学青年教师发展计划经费资助
关键词
微弧氧化
钛合金
神经网络
遗传算法
micro-arc oxidation
titanium alloy
neural network
genetic algorithm
