高速公路交通的神经网络自适应控制

Neural network adaptive control for highway traffic

建立宏观的高速公路交通流动态模型并设计理想的控制方案具有重要的理论意义和实际应用价值.在高速公路交通流宏观模型中含有复杂的非线性关系,一般的最优控制方案在计算上存在维数灾难问题,而且控制目标函数的设计往往有失偏颇.为此,本文建立了一个包括神经网络结构在内的宏观高速公路交通流动态模型,从高速公路管理者和使用者两方面的需求出发,设计了能够反映高速公路总体运营水平的目标函数,并提出相应的自适应控制方案.该方案结构简单、计算方便.数值仿真实验的结果表明控制效果良好,与不施加控制的运营方式相比较能够提高高速公路交通的总体运营水平.

To establish a macroscopical highway traffic flow dynamic model and to design a perfect control scheme it is of great importance for both theory and practice. In the highway traffic flow macroscopical dynamic model there are complicated nonlinear relations, and hence in the general optimization control scheme there may exist the dimension disaster in computing and the design of the control objective function may not be considerate as well. A highway traffic flow macroscopical dynamic model including a neural network structure is established; the objective function that can reflect the total operation level of the highway is then designed according to the demands from both managers and users; and the adaptive control scheme is also given such that it is simple in structure and easy for computation. The numerical simulation shows that the adaptive control scheme given in the paper is favorable: it improves the total operation level of the highway in comparison with that without control.