A novel optimization model based on fuzzy time series for short-term Air Quality Index forecasting

The problem of air quality forecasting has caused a heated discussion among scholars at home and abroad. Short term air quality forecasting substantially impacts regulatory effectiveness to improve environmental and human health. Therefore, many scholars have proposed a lot of Air Quality Index (AQI) forecasting models to improve air quality forecasting performance. However, the fuzziness of air quality data is often ignored, which may reduce the effectiveness of the forecasting. In this study, a new fuzzy forecasting system that includes a fuzzy time series, data preprocessing technique, multi-objective bat optimization algorithm, and forecasting algorithms is proposed to increase air quality forecasting performance and accuracy. To evaluate the forecasting system effectiveness, one-hour AQI data collected from four cities in China are applied in three experiments. The experiment results find that the proposed forecasting system can effectively utilize the fuzziness of air quality data, and it not only improves the forecasting accuracy but also achieves a higher degree of certainty, which can effectively assist air quality supervision.