Clustering of vehicle waveform based on principal component analysis and ART2 neural network

Loop induction detector has been used to collect electromagnetic induction data which were caused by vehicles,which was crossing the toroidal coil under the road.The dimension of vehicle electromagnetic induction data were reduced by principal component analysis and a new sample space was created.The new sample space was employed as the inputs of ART2 network,so the the input dimension of ART2 network was reduced.Through the experimental study,this classification methods can get good results.The construction of ART2 network was simplified,and the convergent speed of the ART2 network was enhanced greatly due to the number of the ART2 inputs was reduced.From the Classification results,a good Accuracy can be got.