摘要
采用图像扫描方法获取碳钢、低合金钢实海挂片的腐蚀形貌图像并进行图像分析;用灰色关联及典型相关技术分析了试片扫描灰度值分布与单位面积平均腐蚀失重及局部腐蚀平均深度的关系;用神经网络理论建立了扫描灰度值分布与试片局部腐蚀平均深度间的关系模型以及扫描灰度值典型加权与试片单位面积平均腐蚀失重间的关系模型.
Scanner is used to acquire corrosion images of carbon steel and low - alloy steel in sea-water and in order to show the corrosion modality clearly the images are pre - processed by the average value filter and non - linear fuzzy enhancement methods. The gray relational analysis and canonical correlation analysis are used to analyze the relations between grey data and corrosion data of metallic samples. The results show that there is higher gray relational grade. The canonical correlation coefficient between grey data and uniform corrosion lost - weight is 0.99 while the coefficient between grey data and localized corrosion depth is 0.98. Using artificial neural network theory, the model between the grey distribution of metallic samples and localized corrosion depth has been developed. According to this model, there are only 3.89 percent absolute error between the predicted result and the real value and the error will decrease if normal samples increase. With a high correlative coefficient, 0. 98, the linear relational model between the grey distribution of metallic samples and uniform corrosion lost - weight has been studied, too.
出处
《中国腐蚀与防护学报》
CAS
CSCD
2001年第6期352-356,共5页
Journal of Chinese Society For Corrosion and Protection
基金
国家自然科学基金重大项目(59899144-3)
关键词
低合金钢
灰色关联
人工神经网络
腐蚀形貌图像
实海挂片
carbon steel, low - alloy steel, gray relational analysis, canonical correlation analysis, artificial neural network