Air quality prediction using optimal neural networks with stochastic variables

Abstract

We apply recent methods in stochastic data analysis for discovering a set of few stochastic variables that represent the relevant information on a multivariate stochastic system, used as input for artificial neural networks models for air quality forecast. We show that using these derived variables as input variables for training the neural networks it is possible to significantly reduce the amount of input variables necessary for the neural network model, without considerably changing the predictive power of the model. The reduced set of variables including these derived variables is therefore proposed as optimal variable set for training neural networks models in forecasting geophysical and weather properties. Finally, we briefly discuss other possible applications of such optimized neural network models.

Prediction of PM10 Concentrations in the Ningdong Base

“…where n i xXR ∈= is input, i yYR ∈= is output, and the aim is to find a real function () fx, then use any input x to infer the corresponding output y [5]. The hyper plane which the desires of classification problems is actually the solution of regression problem.…”

“…By reference [3][4][5] and on-the-spot investigation, the initial concentration (PM 10 0 , ug/m 3 ), wind speed (v, m/s), temperature (T, ℃) and relative humidity (H, %), pressure (p, kPa) , rainfall (R, mm), sunlight (s, H) are selected as input parameters.…”

Variance Characteristics of Tropical Radiosonde Winds Using a Vector-Tensor Method

“…Other more sophisticated methods of analysing such data exist that may also yield further insights into the behaviour of upper-air winds in the monsoonal tropics. For example, artificial intelligence and machine-learning based methods on pattern recognition or trend prediction [19,20] can identify features within the data time series that are not captured by the mean and variance. On the point of time series analysis of stochastic vectors, methods exist by which the time evolution of a random vector can be described.…”

Improving air quality management using gradient boosting based hierarchical temporal memory neural networks and fuzzy based classification based regression tree

“…Thus the proposed system summarizes that the air quality should be monitored in the indoor and in-car system. Ana Russo et al, [8] author uses the stochastic variables to train the artificial neural networks for predicting the air forecasts. The stochastic variable reduces the training time and provide the efficient result when matching the air pollutants in the air forecast.…”