基于动态贝叶斯网络的公路交通事件算法

Detection Methods for Expressway Traffic Events Based on Dynamic Bayesian Networks

利用贝叶斯后验概率函数,通过不断改进有关事件发生概率的权值,充分逼近真实值.其中,对于有关参数数值的获取,我们利用Gibbs抽样,通过随机模拟,即Markov Chain Monte Carlo(MCMC)的方法,来近似得到,尽管是近似,却有很高的精确度.最后,我们用这个方法做了一个交通事件的例子,表明效果很好.

With a method based on Dynamic Bayesian Network for tralfic affairs probability with missing-data and historical data, we continue improving the weights of traffic affairs probability by using Bayesian later estimate equations. To get the true value of parameters, we employ Gibbs sampling method, and Markov Chain Monte Carlo （MCMC） to get random data ,with which to estimate the mean of parameters, and fortunately, this method have high accurate degree. For instance, we employ this method for simulation of expressway tunnel events with good efficiency.