FRAMEWORK APPLICATION FOR THE DEVELOPMENT OF SOFTWARE SYSTEM FOR FORECASTING AIR POLLUTION BY NITROGEN DIOXIDE IN THE INTEGRATED ENVIRONMENT

Abstract

The problem of forecasting air pollution by nitrogen dioxide based on the application of a systems approach is considered in the paper. It is investigated for creating a software system in an integrated environment. Such an environment combines the capabilities of traffic forecasting models with models for forecasting nitrogen dioxide concentration in the air. It strengthens solving the problem by creating a united software system. The need for such integration when both types of models are available and resources are limited is substantiated. It is proposed to use frameworks to develop such a software system, taking into account the existing connections between the components of the system. The software system for forecasting air pollution by nitrogen dioxide has been designed, including determining and integrating the relevant functional capabilities, selecting the appropriate system and database structure. It has been implemented based on creating a web system using the Django web development framework with the integration of relevant Python tools for working with data in an appropriate way. Experimental investigation of the application of the developed system for systematic forecasting of nitrogen dioxide concentration was conducted confirming the effectiveness of the obtained results by reducing the level of forecasting errors based on mean squared error indicator for 6 hour forecasting horizon cumulatively as well as for the sixth hour and increasing the degree of integration of the system components by combining the appropriate set of traffic forecasting models and models for forecasting nitrogen dioxide concentration in the air at different stations due to the reduction in the complexity of the created models as a result. Forecasted traffic values were applied as additional input features for models in the developed software system.