The correlative threshold denoising of wavelet packet transformation is used to process the test data

and the RBF neural network with subtractive clustering and LMS algorithm is used to establish and compensate the transducer model.The study results show that the network structure is simple and astringency is fast

the modeling and compensation by using the new algorithm is effective to correct the nonlinear dynamic character of transducer

and noise elimination with correlative threshold plus wavelet packet transformation is superior to conventional noise elimination methods.