Application of Fuzzy Clustering in Speed Prediction

In order to ensure the effective work of vehicle active safety warning system,vehicle operational state need to be accurately estimated.Vehicle speed most able to characterize vehicle running status can not only reflect driver's operation for pedal,but also reflect the consistency between the current vehicle state and traffic environment.In order to provide reliable basic data for warning system,this paper proposes to predict speed by fuzzy clustering analysis.Speed is discrete into 13 kinds of states according to vehicle dimension.In full consideration of speed observation sequence change and the fuzzy rules,the speed the next moment will be predicted by speed state transition during aperiod.The prediction of speed can be classified with fuzzy logic rules and speed belongs to which state can also be determined.And then,driver's operation can be determined through the prediction results of speed.Finally,the real testing data were used to validate the proposed algorithm and the test results show that this algorithm can accurately predict speed within short time.