全路网改进动态O-D反推模型

针对动态O-D矩阵在全路网中难以直接获得的问题,提出了一种改进的反推模型。首先提出了以最小化观测值与反推值偏差的绝对值之和作为目标函数的路口参数优化模型,并采用遗传算法求解,设计了编解码方案。将反推得到的路口转向流量和流量检测系统中得到的路段流量共同作为已知量,建立了其与全路网动态O-D矩阵的动态关系,增强了系统的静定性。以反推值与最优历史值的偏差作为状态变量,建立了基于Kalman滤波的状态空间模型,并采用扩展Kalman滤波求解。仿真结果表明,模型和算法具有较好的精度、效率和鲁棒性。

Estimates of dynamic origin-destination flows for general road networks

Dynamic origin-destination (O-D) flows in general road networks are very difficult to obtain directly. A parameter optimization model was developed to estimate intersection turning flows with an objective function that minimizes the sum of the absolute deviations between observed and estimated values. The model solution is obtained using a genetic algorithm with appropriate encoding and decoding methods. After the estimated intersection turning flows are correlated with the link counts from traffic surveillance systems, the dynamic interrelations between the turning flows and the O-D flows are formulated, which improves system observability. The O-D flow deviations from the optimal historical data are used as state variables in a revised state-space model based on Kalman filtering which is solved using extended Kalman filtering. The simulation results show that the approach is quite accurate, efficient, and robust.