基于多元数据融合的生态环境污染物类型识别研究  

Study on Type Identification of Eco-environmental Pollutants Based on Multivariate Data Fusion

作　　者：李鹿珍 Li Luzhen(Liaoning University,Shenyang 110024,China)

机构地区：[1]辽宁大学,辽宁沈阳110024

出　　处：《环境科学与管理》2022年第10期73-77,共5页Environmental Science and Management

摘　　要：为提出更有针对性的生态治理措施,文章基于多元数据融合,研究一种生态环境污染物类型识别方法。先利用环保在线监测数据采集仪采集多元污染数据,并实施去量纲、缺失数据填补、去噪等处理。利用PCA方法从数据中选取特征,以特征为输入,通过神经网络算法确定每种单一特征对应下的决策结果。利用D-S证据理论对决策结果实施融合,最终识别出对应的生态环境污染物类型。结果表明:应用所研究方法后,可明确检测出水源中的污染类型,即水源3存在无机主导型污染问题,水源4存在营养主导型的污染问题。In order to propose more targeted ecological governance measures, this time, based on multivariate data fusion, a method for identifying types of ecological environmental pollutants was studied. First, use the environmental protection online monitoring data acquisition instrument to collect multi-pollution data, and implement de-dimensioning, missing data filling, denoising and other processing. The PCA method is used to select features from the data, and the feature is used as the input to determine the decision result corresponding to each single feature through the neural network algorithm. The D-S evidence theory is used to fuse the decision results, and finally identify the corresponding types of ecological environmental pollutants. The results show that after the application of the research method, the types of pollution in the water source can be clearly detected, that is, the water source 3 has the inorganic-dominated pollution problem, and the water source 4 has the nutrient-dominated pollution problem.

关 键 词：多元数据融合 污染物类型 PCA方法 神经网络 D-S证据理论 

分 类 号：X171[环境科学与工程—环境科学]