|
Abstract:
|
O congestionamento urbano tem crescido notoriamente e os impactos negativos causados s#ao os mais diversos, transitando do aumento de tempo de jornada a quest#oes de sa´ude e ambientais. Uma alternativa no combate a esse efeito se d´a atrav´es do aumento da efici#encia das malhas vi´arias urbanas. Diversas estrat´egias de controle em tempo-real existem com esse intuito, sendo a Traffic-responsive Urban Control , ou TUC, de especial interesse por apresentar bons resultados pr´aticos e estar consolidando-se em grandes centros. Essa, por´em, efetua o controle de fra¸c#oes de verde de forma heur´ýstica, atrav´es de um regulador quadr´atico-linear de dois est´agios. Prop#oe-se um novo m´odulo de controle para este fim utilizando a teoria de controle preditivo, o qual considera as restri¸c#oes intr´ýnsecas do modelo de redes vi´arias explicitamente, garantindo a otimalidade do sinal de controle calculado para o horizonte de tempo observado. Posteriormente modifica-se o m´odulo de controle para que opere de forma distribu´ýda, estabelecendo condi¸c#oes de converg#encia `a solu¸c#ao ´otima. Resultados num´ericos e de simula¸c#ao da aplica¸c#ao do m´etodo a duas malhas vi´arias modelos s#ao apresentados e demonstram um desempenho ora equivalente, ora superior ao regulador original da estrat´egia TUC. Urban congestion has grown notoriously and the negative impacts range from increased journey times to health and environmental issues. A way to cope with this effect is to increase urban network efficiency. For this purpose several real-time control strategies have been developed, one being the Traffic-responsive Urban Control, or simply TUC. This approach has consolidated itself among the representative real-time control strategies and is of particular interest for its good performance in practical emplementations. Nevertheless, TUC uses a heuristic two-stage linear-quadratic regulator to perform split control. A new split control module is proposed to replace the original sub-optimal one. Based on model predictive control, this new strategy can handle the intrinsic constraints of the urban network model explicitly, therefore ensuring optimality of the solution found for the given prediction horizon. A distributed version of the proposed control is developed, along with conditions for its convergence to the centralized optimal solution. Numerical results and simulations from the application of the methods to two urban networks show that the predictive control strategies usually produce a performance better than or at least comparable to the baseline TUC regulator. |
