自适应单指数平滑法在短期交通流预测中的应用

短时交通流预测是实现交通规划和管理的关键技术之一. 指数平滑法因其计算过程简单, 需要观测数据较少等优点, 在短时交通流预测中获得了广泛的应用, 但其平滑系数缺乏有效的选取方法. 本文提出了一种自适应单指数平滑法, 通过近似动态规划方法的引入, 结合实际交通流数据对指数平滑系数进行优化, 使其随预测过程自动更新, 从而保证了预测的实时性、准确性. 严格的理论推导证明了这种预测方法的收敛性, 仿真结果验证了算法的有效性.

Application of adaptive single-exponent smoothing for short-term traffic flow prediction

Short-term traffic flow prediction is a key technique to realize the transportation planning and management. Because of the simple calculation and the small number of observation data, the exponential smoothing is extensively applied as an important forecast method. However, in the traditional method, there is no theoretical method for selecting the smoothing coefficient. We propose an adaptive single-exponent smoothing approach to optimize the smoothing coefficient automatically based on the approximate dynamic programming. With rigorous analysis, it is shown that the proposed prediction scheme guarantees the convergence. The simulation results validate the effectiveness of the proposed algorithm.