Prediction of particular matter concentrations by developed feed-forward neural network with rolling mechanism and gray model
- 616 Downloads
Particular matter (PM) due to its side effects on human health like increase the risk of lung cancer and vision impairment has been one of the major concerns for air quality. These particles are now considered as one of the high priorities issues by health organizations in China. In this study, daily PM2.5 and PM10 concentrations data from November 2013 to January 2014 in Hangzhou, Shanghai and Nanjing (three important cities in Yangtze River delta of China) were used to introduce more suitable method to forecast air PM2.5 and PM10 concentrations. Feed-forward neural networks (FFNN) have been introduced as a possible forecasting model for complex air quality prediction. However, due to its deficiency to assess the possible correlation between different input variables, an enhanced FFNN with rolling mechanism (RM) and accumulated generating operation (AGO) of gray model (RM-GM-FFNN) was developed. RM and AGO were used to address the trends of input samples of FFNN and detract the randomness of the input data of FFNN, respectively. Both FFNN and RM-GM-FFNN were tested for prediction of the daily PM2.5 and PM10 concentrations with meteorological parameters and historical PM concentration during the given time. The numerical results showed that in all cases, the coefficient of determination (R 2) and the index of agreement of RM-GM-FFNN increased, while the root-mean-square error (RMSE) and mean absolute error of RM-GM-FFNN decreased. In addition, the mean bias error was more close to zero when compared with that of FFNN, indicating that RM-GM-FFNN performed a better accuracy.
KeywordsFeed-forward neural network Gray model Particular matter concentration Rolling mechanism
This work was financially supported by the public industrial technology research project of Zhejiang province, China (No. 2013C31075).
- 19.Zheng H, Shang X (2013) Study on prediction of atmospheric PM2.5 based on RBF neural network. In: 2013 Fourth International Conference on Digital Manufacturing and Automation (ICDMA), Qingdao, pp 1287–1289. doi: 10.1109/ICDMA.2013.306
- 25.http://www.tianqihoubao.com/aqi/. Accessed 8 Aug 2014
- 26.http://www.wunderground.com/history/airport/ZSNJ/2013/12/1/DailyHistory.html. Accessed 8 Aug 2014
- 27.Díaz-Robles LA, Ortega JC, Fu JS, Reed GD, Chow JC, Watson JG, Moncada-Herrera JA (2008) A hybrid ARIMA and artificial neural networks model to forecast particulate matter in urban areas: the case of Temuco, Chile. Atmos Environ 42:8331–8340. doi: 10.1016/j.atmosenv.2008.07.020 CrossRefGoogle Scholar
- 30.Voukantsis D, Karatzas K, Kukkonen J, Räsänen T, Karppinen A, Kolehmainen M (2011) Intercomparison of air quality data using principal component analysis, and forecasting of PM10 and PM2.5 concentrations using artificial neural networks, in Thessaloniki and Helsinki. Sci Total Environ 409:1266–1276. doi: 10.1016/j.scitotenv.2010.12.039 CrossRefGoogle Scholar