Intelligent signal-noise separation based on a support vector machine and denoising method library for massive-scale induced polarization data

Dear,
This code package is an example to show the training and testing process of support vector machine (SVM) noise classifier and denoising method library.
AA_main1_signal.m:  This main program is used to generate induce polarization pure signals.
AA_main2_noise.m:  This main program is used to add different types of noise interference combinations to the IP signal and extract statistical features.Finally, two input-output sample sets are generated for training and testing respectively.
AA_main3_train_combined.m: This main program is used to train the SVM noise classifier model based on input-output samples.
AA_main4_test_combined1.m: The main program is used to test the accuracy of noise recognition.
AA_main4_test_combined2.m: The main program is used to select a group of polluted signals to identify noise and call the signal processing technology in the de-nosing library to separate the interference.
AA_main4_test_combined3.m: The main program is used to test the effectiveness of noise identification and separation using measured IP signals. Finally, constrained parameter estimation is carried out for the signal after noise reduction, and cole-Cole model parameters are extracted.
AA_main4_test_combined4.m: The main program is used to identify and separate signal noise from 10,000 sets of simulation data.The signal errors before and after noise reduction are calculated respectively.

In the code above, these programs (AA_main1_signal.m, AA_main2_noise.m, AA_main3_train_combined.m, AA_main4_test_combined4.m) take a long time to compute, and you can skip them. You can test the algorithm by running these pieces of code  (AA_main4_test_combined1.m, AA_main4_test_combined2.m, AA_main4_test_combined3.m)  directly.

The above codes have been successfully run on MATLAB R2018a.