A collaborative predictive multi-agent system for forecasting carbon emissions related to energy consumption.

Energy production and consumption are one of the largest sources of greenhouse gases (GHG), along with industry, and is one of the highest causes of global warming. Forecasting the environmental cost of energy production is necessary for better decision making and easing the switch to cleaner energy systems in order to reduce air pollution. This paper describes a hybrid approach based on Artificial Neural Networks (ANN) and an agent-based architecture for forecasting carbon dioxide (CO2) issued from different energy sources in the city of Annaba using real data. The system consists of multiple autonomous agents, divided into two types: firstly, forecasting agents, which forecast the production of a particular type of energy using the ANN models; secondly, core agents that perform other essential functionalities such as calculating the equivalent CO2 emissions and controlling the simulation. The development is based on Algerian gas and electricity data provided by the national energy company. The simulation consists firstly of forecasting energy production using the forecasting agents and calculating the equivalent emitted CO2. Secondly, a dedicated agent calculates the total CO2 emitted from all the available sources. It then computes the benefits of using renewable energy sources as an alternative way to meet the electric load in terms of emission mitigation and economizing natural gas consumption. The forecasting models showed satisfying results, and the simulation scenario showed that using renewable energy can help reduce the emissions by 369 tons of CO2 (3%) per day. [ABSTRACT FROM AUTHOR]