关键词: 事件检测, 模糊C均值聚类, 粗糙集, 模糊推理, 属性离散化

Abstract: In order to detect traffic incident accurately，reduce traffic delays and ensure road safety，a new detection model using FCM and fuzzy rough sets is presented.The model is composed by three parts such as discretization，reasoning rules establishment and fuzzy reasoning.A new method is proposed when attribute discretization.Using custom membership function to fit results of the FCM clustering，the membership function and parameters are obtained to discrete the attribute data without clustering when new data come.Then reasoning rules are formed using the rough set theory so as to improve fuzzy reasoning.Finally Max-Min fuzzy reasoning method is adopted to incident detection.By comparative testing，the better performance is achieved using new model and result valid the validity of new model which provides a new idea for automatic incident detection.

Key words: incident detection, fuzzy C mean clustering, rough sets, fuzzy inference, attribute discretization