基于HMM的电动出租汽车行驶状态预测改进模型研究

IMPROVED PREDICTION MODEL OF ELECTRIC TAXI DRIVING STATE BASED ON HMM

对电动出租汽车行驶状态预测是交通状况和负荷预测方面的重要研究内容。通过模拟电动出租汽车行驶状态,匹配电动出租汽车特殊的行驶特征,提出一种基于隐马尔可夫模型(HMM)的行驶状态预测改进模型。利用出租汽车GPS行驶数据,通过载客情况和停留识别算法进行行程划分。在求解电动出租汽车的行驶状态时,使用滑动窗口模型改进状态转移概率求解,通过Baum-Welch算法求解观察概率与初始概率分布。测试结果表明HMM能准确地对电动出租汽车行程的目的地与行驶里程进行预测。

The driving state prediction of electric taxi is an important research content in traffic condition and load prediction.By simulating the driving state of electric taxis and matching the special driving characteristics of electric taxis,an improved model of driving state prediction based on hidden Markov model(HMM)is proposed.Using the taxi GPS driving data,the itinerary was divided by the passenger situation and the stay identification algorithm.When solving the driving state of the electric taxi,the sliding window model was used to improve the state transition probability solution,and the Baum-Welch algorithm was used to solve the observation probability and the initial probability distribution.The test results show that the HMM model can accurately predict the destination and mileage of the electric taxi trip.