Hybrid and Reinforcement Multi Agent Technology for Real Time Air Pollution Monitoring

Abstract : This paper describes the design and implementation of a modular hybrid intelligent model and system, for monitoring and forecasting of air pollution in major urban centers. It is based on Multiagent technologies, Artificial Neural Networks (ANN), Fuzzy Rule Based sub-systems and it uses a Reinforcement learning approach. A multi level architecture with a high number of agent types was employed. Multiagent’s System modular and distributed nature, allows it’s interconnection with existing systems and it reduces its functional cost, allowing its extension by incorporating decision functions and real time imposing actions capabilities.