短时交通流预测FSMSVR模型

针对城市道路短时交通流的复杂非线性特点和以往的预测仅考虑典型交通条件（无交通事故等突发事件）的现状，结合交通流的特征，提出了一种有限状态机支持向量回归模型（finite state machine of support vector regression model，FSMSVR）的短时交通流预测机制. 通过线性回归算法和指数平滑算法划分交通流状态，根据各状态特点结合支持向量回归算法建立有限状态机工作机制，实现涵盖典型和非典型交通条件的短时交通流预测. 通过实验例证，对比了FSMSVR模型和传统SVR模型对城市道路6min交通流的预测，研究结果表明，该预测机制能够提高预测精确度，在非典型条件下有着较好的预测表现.

A FSMSVR model of short-term traffic forecasting

Short-term traffic flow of urban roads is complex and nonlinear, while most existing literature about short-term traffic flow forecasting only consider traffic flow under normal conditions which do not include incident. So a new forecasting algorithm, finite state machine of support vector regression model, is put forward. We divide traffic flow state by linear regression and exponential smoothing, establish finite state machine base on SVR and traffic flow state, and make the method is able to adapt to both normal and the atypical conditions. An empirical study, contrasting the FSMSVR model to traditional SVR model about traffic flow prediction in 6 min, shows that the forecast mechanism can increase the prediction accuracy and has a good performance on the prediction of atypical conditions.