Emotional temporal difference Q-learning signals in multi-agent system cooperation: real case studies

Chaotic non-linear dynamics approach is now the most powerful tool for scientists to deal with complexities in real cases; and artificial neural networks and neuro-fuzzy models are widely used for their capabilities in non-linear modelling of chaotic systems. Chaos, uncertain behaviours, demanding fluctuation, complexity of the traffic flow situations and the problems with those methods, however, caused the forecasting traffic flow values to lack robustness and precision. In this study, the traffic flow forecasting is analysed by emotional concepts and multi-agent systems (MASs) points of view as a new method. Its architecture is based on a temporal difference (TD) Q-learning with a neuro-fuzzy structure. The performance of TD Q-learning method is improved by emotional learning. The concept of emotional TD Q-learning method is discussed for the first time in this study. The forecasting algorithm which uses the Q-learning algorithm is capable of finding the optimal forecasting approach as the one obtained by the reinforcement learning. In addition, in order to study in a more practical situation, the neuro-fuzzy behaviours can be modelled by MAS. The real traffic flow signals used for fitting the proposed methods are obtained from interstate I-494 in Minnesota City in USA and the E17 motorway Gent-Antwerp in Belgium.