基于线性变换的水质综合评价方法

对于多个因子的水质综合评价存在很多不同方法,尚未形成公认的权威评价方法。因而,对水质综合评价方法进行探索有利于积累研究经验,丰富方法手段。在把数据向0 1]区间标准化的基础上,通过线性空间变换,把水质样本数据与《地表水环境质量标准》(GB3838-2002)数据转换到同一线性空间。计算水质样本数据向量与5类水质划分等级标准数据向量的欧式距离,把欧式距离的最小值作为水质类别辨别的依据,最终得到水质的综合评价结果。借由已经公开发表的19个断面水质监测数据,运用大型工程计算软件Matlab 2010b进行计算,并与变权欧式距离模型、灰色聚类法、模糊综合指数法和BP神经网络法的评价结果相比较,符合性较好,验证了线性空间变换法应用于水质综合评价的科学性和合理性。方法可用于水质综合评价。在不对污染因子进行加权的情况下,基于线性变换的水质综合评价方法,获得的评价结果较轻。对于被评价为同一类水质的不同监测断面,可以借助"评价单元与5级水质标准的欧式距离矩阵",对评价单元水质差异进一步辨识。

COMPREHENSIVE ASSESSMENT OF SURFACE WATER QUALITY BASED ON LINEAR TRANSFORMATION

Comprehensive assessment of water quality is an important basis for the calculation of water environmental capacity and the implementation of water pollution control in water environment monitoring. Many different methods employ multiple factors to assess the comprehensive conditions of water quality, but no common consensus has been reached. Therefore, the study of water quality assessment using multiple factors could facilitate the comprehensive assessment of water quality. The linear transformation based on Euclidean distance was adopted to categorize the multiple water quality variables in this study. Firstly, both the raw monitor data and five grades surface water environment quality standards (GB3838-2002) were standardized, then the standardized variables were transformed to the same linear space using Matlab 2010b. The minimum Euclidean distance between vectors of water quality and the five standards in the linear space were used to identify the different categories of water quality variables. Using the published water quality monitoring data in 19 sections, including water quality monitoring data of Qiantang River tributaries, East Village section of Yangjiang, Xu, and Panlong River, and 4 Wells monitoring data of XianYang City, and JingYang County. Linear space transformation method was applied to comprehensive evaluation of water quality on each section. We used Matlab 2012b to perform calculations. Our results indicated that the linear space transformation based on minimum Euclidean method was suitable for the comprehensive assessment of water quality, and compatible with other analytical methods, such as varying weights continental distance model, grey clustering, fuzzy comprehensive index, and BP neural network. In the case of no pollution factor weighted, the evaluation result of water quality comprehensive evaluation based on linear transformation method is lighter than result base on factor weighted methods. However, even under the condition that water quality is very good, we can also use “Euclidean distance matrix” to recognize the difference between water quality monitoring data. In addition, under the condition of keeping evaluation methods unchanged, the linear space transformation method was suitable for the improved water quality standard of surface water, in the future. Without weighted pollution factors, the linear space transformation method can also compatible with current approaches in a simplified manner. At the meantime, the pollution and other monitor factors in standardization all can be treated as one direction positive/negative factors, not two directions needed. This can also facilitate data processing in water quality assessment