| 石雅楠,付冬梅,支元杰,陈闽东.基于ANFIS改进的大气腐蚀环境缺失数据填补方法[J].装备环境工程,2016,13(6):78-84. 石雅楠,付冬梅,支元杰,陈闽东.A Imputation Method Based on Improved ANFIS for Environmental Data[J].Equipment Environmental Engineering,2016,13(6):78-84. |
| 基于ANFIS改进的大气腐蚀环境缺失数据填补方法 |
| A Imputation Method Based on Improved ANFIS for Environmental Data |
| 投稿时间：2016-04-12 修订日期：2016-11-09 |
| DOI：10.7643/ issn.1672-9242.2016.06.014 |
| 中文关键词: 大气腐蚀 缺失数据 填补 相关因素 ANFIS |
| 英文关键词:atmospheric corrosion missing data imputation Relevance Factors ANFIS |
| 基金项目:国家重点基础研究发展计划（973计划） |
|
|
|
|
| 摘要点击次数: |
| 全文下载次数: |
| 中文摘要: |
| 目的 针对大气腐蚀中重要环境数据缺失的复杂问题，提出一种相关因素(Relevance Factors)和自适应神经模糊推理系统(Adaptive Neuro-Fuzzy Inference System)结合的方法(RF-ANFIS)对缺失数据进行填补。方法 首先采用相关因素方法计算缺失数据和多项环境因素间的相关程度，筛选出相关系数较大的因子，然后应用ANFIS构建缺失数据与所选环境因子的关系模型。最后以二氧化硫数据为具体对象，采用北京2015年的气象数据对所建立的模型进行检验。结果 经过改进的RF-ANFIS模型在最优情况下样本均方误差为0.696，在14个测试样本中13个相对误差在20%以内，针对有限样本的数据分析中更为适用。结论 该方法有效提高了大气腐蚀环境数据缺失的填补精度，对在数据缺失情况下预测大气腐蚀速率具有重要意义。 |
| 英文摘要: |
| Objective A new method based on Relevance Factors and Adaptive Neuro-Fuzzy Inference System was proposed to impute missing important environmental data of atmospheric corrosion. Methods Through Relevance Factors, the relevance degree between missing data and a number of environmental factors was calculated, selecting valuable factors with high relevance?coefficient, then the relationship model between missing data and chosen factors could be built through ANFIS. Finally taking the imputation of missing SO2 data as the object, the RF-ANFIS model was tested by atmospheric data of Beijing in 2015. Results The RF-ANFIS model were much better than BP in?sample?distribution and error evaluation, which was feasible in finite sample analysis. Conclusion The new method efficiently improves the accuracy of imputing environmental data in atmospheric corrosion, which is vital to predict atmospheric corrosion rate with missing data. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
