基于隐马尔可夫模型和遗传算法的地图匹配算法

Map Matching Algorithm Based On Hidden Markov Model and Genetic Algorithm

综合采用隐马尔可夫模型(HMM)和遗传算法,提出了一种新的地图匹配算法.首先初始化HMM概率矩阵,然后使用前向后向算法进行参数学习,用Viterbi算法预测一组路段序列,最后将路段序列作为种群,通过遗传算法得到最优的路段序列.采用北京市2012年出租车GPS定位数据分别对传统的基于隐马尔可夫模型的算法和新算法进行测试,实验结果表明,传统的基于隐马尔可夫模型的算法的匹配精确度低于90%,新算法的匹配精确度高达90%以上.

A new map matching algorithm was proposed on the basis of the hidden Markov model Then,the sections Firstly,the using and the genetic algorithm.learned HMM probability matrix was initialized.algorithm,and parameters were by the forward-backward a set of road was population,the Finally,taking using predicted optimal by using the algorithm.section sequence as Viterbi section sequence 2012 was obtained by the genetic algorithm.By using the taxi GPS data from Beijing in to test the traditional algorithm the based on hidden Markov based model hidden and the algorithm,the proposed model results showed below that traditional the algorithm on Markov has a matching accuracy 90% and proposed algorithm has a matching accuracy above 90%.