The quick analysis system for underground pipeline coating detection has been set up.Defect orientation of the system is
made by using the dyadic wavelet transform of ON,instant OFF
close interval potential survey(CIPS) potential and the difference of them.It is indicated that the extreme values of detail-coefficients of dyadic wavelet transform will increase with the scale at the damaged points otherwise it will decrease with the scale at the other places.With the product of the third layer detail-coefficients and the product of the third layer smoothed-coefficients of those three curves obtained,the damaged points of the coating could be easily found.To diagnose the coating state of the system the wavelet Kohonen
neural network is set up.The front layers of the network are corresponding to multiresolution decompositions,which has the ability of picking up the information.The last layer of the model is corresponding to Kohonen neural network,which has the ability of self-training.The coating state can be quickly diagnosed after the galvanostatic transient response is input to the model.The CIPS is used in pilot detection.Then the damaged points of the coating can be made certain by the defect orientation part of the system.The galvanostatic transient response method is used to detect the coating state of buried pipelines at the damaged point.Then the coating state can be judged by the coating state diagnosis part of the system.The detection result of real pipeline between Dagang and Cangzhou of Tianjin Dagang oil field was satisfactorily.