基于多源信息融合决策的灌区生态环境评价指标优选

Index optimization of eco-environment evaluation in irrigation district based on multi-source information fusion decision

针对目前灌区生态环境评价研究中指标选取主观化、未成体系化的困境,该研究提出了一种基于指标多源特征信息融合的优选思路与方法。以关中地区泾惠渠灌区为研究对象,通过构建灌区生态环境概念模型,以"灌区资源"定义因素层,逐级确定了31个指标,初步构建了灌区生态环境评价指标库。采集指标灵敏度、均衡度、独立性这3种特征信息,视为3项证据,进而采用DS(Dempster Shafer)证据理论对该3项证据基本信任分配函数值进行融合,计算该3项证据共同支持作用下的指标可信度,依据可信度高低排序结果,即可从繁复、庞大的指标库中进行指标优选。该研究设置了两种融合思路:1指标全体融合、2先分别剔除3项证据下基本信任分配函数值较低的指标,然后对剩余指标进行融合;并进行对比分析研究,结果表明:第2种思路下的计算会提前剔除某些可信度较高的指标,如盐碱化程度、空气相对湿度指标;按照第1种思路,对指标全体进行信息融合,进而根据可信度排序结果,选取可信度较高的指标作为灌区生态环境评价研究的主评指标,为灌区生态环境健康评价提供了支撑。为其他领域评价研究中指标的优选提供了新的思路和决策方法。

At present, there have been no definite ways and applicable decision-making methods on the optimization of evaluation indexes, especially on the dilemma of the subjectification and non-systematization of the index selections in the ecological environment evaluation researches. Based on the index multi-information fusion, an optimized thought and a decision-making method has been put forward in this research. After collecting the three categories of feature information of index sensitivity, equilibrium degree and independence, the method of orthogonal experiment was used to calculate the index sensitivity in purpose of expressing the dynamic characteristics of indexes. The index equilibrium degree with the help of the definition of Gini coefficient in mathematical economics was calculated to select the simplest indexes that cover the most properties of the evaluation object. The independence of the indexes by means of index correlation analysis was determined to exclude some indexes with high degree of association. After that, DS （Dempster Shafer） evidence theory was adopted for the information fusion. The mentioned three index features can be regarded as three evidences, with the integration of the basic trust distribution function value. The index credibility jointly supported by the three evidences and optimized from the complicated and large index library according to the sequence results of the credibility. This research took the Jinghuiqu Irrigation District as the researching object. To understand connotation of ecological environment in irrigated district from perspective of system, the ecological environment concept model of the irrigation district was constructed to express driving factors of ecological environment in irrigated district, information carrier and representation state. The directed line segment was used to describe interaction and influence relation between different carriers, and the information expressed by this model was sorted. The factor layer was defined as ＆quot;resources in the irrigation district＆quot;, and 31 indexes gradually determined. The ecological environmental evaluation index library of the irrigation district was primarily constructed. The DS evidence theory was used to realize the multi-source information fusion of the basic trust distribution function value of the three evidences. Two fusion thoughts were set：①index entire fusion,②separately remove the indexes of lower basic trust distribution function values under the three evidences, and then carry out fusion to the left indexes. The sequence of credibility of the index feature information was carried out after fusion from high to low, and the calculation results by the above two thoughts were compared and analyzed . The results showed that： the calculation under the thought②would remove some indexes with relatively high credibility in advance, such as degree of acid and alkali, air relative humidity indicator. According to the thought①, results for indexes credibility were ranked as follows： multiple cropping index ＆gt; groundwater depth ＆gt; solid erosion modulus ＆gt; development and utilization rate of surface water resources ＆gt; intensity of pesticide application ＆gt; biodiversity index ＆gt; vegetation coverage ＆gt; intensity of chemical fertilizer＆gt;underwater mineralization＆gt;times of the heavy metal content of soil exceeding the background values ＆gt; species richness ＆gt; air relative humidity ＆gt; overgrazing index ＆gt; drought index ＆gt; salinization index ＆gt; drainage engineering coverage ＆gt; rate of channel lining ＆gt; biomass ＆gt; urbanization rate ＆gt; standard sewage discharge rate ＆gt; surface water NH3-N concentration ＆gt; utilization coefficient of irrigation water ＆gt; percentage of impervious area ＆gt; density distribution of wells ＆gt; surface engineering of water supply capacity. Indexes with relatively high credibility could be selected as main indexes for research on ecological environment evaluation in irrigated area, which provided new thought and policy as well as method with optimized indexes for research on ecological environment evaluation in irrigated district as well as evaluation research in other fields.