Air quality evaluation method based on progressive DS evidence theory

  Objective  Evaluating air quality scientifically is crucial for prevention and control of air pollution. Fuzzy synthetic evaluation (FSE) method has been widely used to comprehensively evaluate air quality, considering the influence of multiple pollutants. But the air quality level evaluated is often lower than the national air quality index (AQI) level, which weakens the impact of multiple primary pollutants on air quality. DS evidence theory has great advantages in uncertainty or ambiguity issues, such as the air quality levels. However, there are few studies to comprehensively evaluate air quality using DS evidence theory. Basic probability assignment (BPA) and evidence conflicts are key issues in DS evidence theory, which may lead to counterintuitive results.

  Method  Therefore, in order to establish a highly reliable BPA, Pearson and Spearman correlation coefficients were used to analyze the air pollutant concentration data. And a progressive DS evidence theory strategy (Pro-DS) was proposed to avoid the conflict of evidence fusion. Thereby, a comprehensive evaluation model of ambient air quality based on Pro-DS was established. Then, the Pro-DS model was applied to Tianjin’s air quality evaluation in 2019. Finally, this paper proposes the comprehensive pollution relative coefficient (CPRC) to quantify the air quality under the action of multiple pollutants.

  Result  Taking the AQI level as an indicator, the F1-score of Pro-DS method was 4.58% to 27.46% higher than that of the widely used models based on FSE, which validated the superiority of Pro-DS model. Since the AQI standard was determined by a single pollutant, the F1-score of all FSE models was lower than 50%. However, taking the CPRC as an indicator, the F1-score of all FSE models exceeded 89.0%, which validated the effectiveness of CPRC. Experimental results showed that Pro-DS model had an F1 score of 93.1%, superior to other widely used FSE models.

  Conclusion  The comprehensive air quality evaluated by Pro-DS model is lower or higher than the air quality evaluated by AQI, which better reflects the comprehensive effect of multiple pollutants on air quality. In this paper, the Pro-DS model can comprehensively evaluate the air quality, and the CPRC indicator can rank the degree of daily pollution, which can provide a substantial reference for the construction of ecological environment projects.