An Ozone Prediction in Atmosphere in Suzhou Based on Deep Learning

Ozone is a common component in atmosphere, but as the ground level atmosphere pollution increases, the Ozone soon becomes a pollutant at ground level atmosphere. In the essay, data (of other pollutants) from a single weather station in Suzhou will be utilized and two different models will be used to predict (single hour) and forecast (12–72 h) ozone levels in atmosphere. Also, different parameters will be tested to evaluate their impact on the model. The single-hour prediction can reach R scores as high as 0.9, RMSE 7.826 μg/m3, and it is based on LSTM, where the 72-h forecast is based on DNN, and has a R score up to 0.4, test RMSE reaching 26 μg/m3.