抄録
For forecasting pollutant concentrations in urban area, a two-stage method is common. First, vehicular emissions are calculated; second, levels of pollutant concentrations are estimated. This approach, however, is not sufficient to fully analyze complex cause-effect relationships between pollutant concentrations and underlying factors. An alternative introduced recently is a neural network approach, which can analyze the complex relationships by approximating input-output responses, and mainly applied to forecasting short-term pollutant concentrations for environmental monitoring and management. However, it requires much work and time to find an optimal solution and optimal network structure. This paper, therefore, proposes a new evolutionary neural network method based on genetic optimization by incorporating genetically optimized step size, momentum and number of hidden units. The proposed method is applied to analyze levels of pollutant concentrations in Dalian City, China. The results show that the proposed method can contribute to improve model estimation and interpretability of neural network approaches.
