金属土壤腐蚀模型中关键参数选择的研究

研究了腐蚀因素的分类以及腐蚀模型中关键参数的选择等问题.将原始数据转换成新的相关系数矩阵,用模式识别方法显示腐蚀因素的分类特性图.为成功建立腐蚀模型,关键参数组合中的腐蚀因素选择不仅要考虑因素的重要性(权重),还应从分类图中按不同组或区域来提取.作为实例,对塔里木地区土壤腐蚀数据集进行了处理,全部24种腐蚀因素散布在因素分类图中,大部分腐蚀因素聚集四个组,即:“含盐量”,“电阻率”,“土质/含水”和“缓冲性”等.对选择最佳关键参数组合的准则进行了讨论,并用一些区域土壤腐蚀模型的工作进行验证

A STUDY ON SELECTION OF KEY FACTORS IN SOIL CORROSION MODELING

In this paper the classification of corrosion factors and the selection of key parameters in corrosion modeling were studied. Based on the correlation coefficient matrix recreated from original data set, the classification diagram of corrosion factors was displayed by means of Pattern Recognition(PR). For creating corrosion modeling successfully, it is better to choose factors both from the important factors with big weight and from different cluster or location of the diagram. As a sample a data set with 24 corrosion factors from soil investigation in Talim area was treated in this way. All corrosion factors were scattered widely in the diagram and most of them gathered into four clusters, namely, "salts", "resistance", "soil texture/water", and "buffer capacity". The criterion of the selection of optimum key parameter set was also discussed and confirmed by several works of regional corrosion modeling.