行驶工况, 构建, 马尔科夫过程, 聚类

Two factors(the mean speed and speed deviation)with the fuels of automobiles were extracted by the factor analysis, according to two factors a series of driving cycles were divided into 3 different clusters by the two-step clustering method. The first cluster represented driving patterns of moving vehicles under 20 km/h.The second cluster consisted mostly of data with average speeds between 20 to 40km/h. And the last cluster included data with average speeds greater than 40km/h. Every driving cycle was taken as a stochastic Markov process with time and the driving data for each cluster were divided into acceleration segments, deceleration segments, idling segments and cruise segments according to a new classification scheme and the transition probabilities of segments were calculated. According to this basis the selection method of candidate cycles was considered by combination the transition probabilities of segments and vehicle specific power of the engine, the evaluation criteria of 12 characteristic parameters of evaluating the representative driving cycle were established. The driving cycles for different clusters were constructed with an example of typical roads in Hefei. The results show that the driving cycles with emissions have high similarity with the experiments and accords with driving cycles of real roads.

driving cycle, construction, Markov process, cluster