Abstract: Although previous clustering algorithms can reduce the communication cost between moving objects and central database in road traffic network, the clustering granularity is set by experiences. hhis paper analysed the influence factors on clustering distance granularity, and introduced a novel method to train historical data with I3P network, and then got clustering distance granularity and clustering time granularity dynamically. Being new historical data, these granularity values can be made to train BP network further. This network can self-adapt in respect of influence factors dynamically, and birth efficient clustering granularity values to reduce communication cost, and forecast traffic jams as optimal route planning's observation.

Key words: BP network, Self-adaptable granularity, Road-traffic network, Moving objects, Clustering algorithm