Seismic Random Noise Suppression Using Optimal ANFIS as an Adaptive Self-Tuning Filter and Wavelet Thresholding

Random noise attenuation plays a vital step in seismic signal processing. Numerous attenuation algorithms have been developed to separate and remove the random noise; nevertheless, they have failed to attain high precision. In this work, a hybrid framework based on an optimal adaptive neuro-fuzzy inference system (OANFIS) and a recent wavelet thresholding (WT), specifically OANFIS WT, is proposed to attenuate the random noise present in the seismic signals. In the suggested OANFIS WT method, the OANFIS extract the relevant seismic signal information from the contaminated signal using the premise and consequence parameters of ANFIS. These parameters are determined optimally using the Honey badger algorithm with mean square error value as an objective function. Here, OANFIS acts as an adaptive self-tuning filter that extracts the appropriate seismic signal information without any knowledge of the amount of noise in the contaminated signal. Therefore, some noise may be present in the output of OANFIS. Thus, the WT is applied to the extracted signal, with different values of the adjusting parameters in the thresholding function, to attenuate the noise effectually. Lastly, the experimental results on the synthetic and real seismic signals reveal that the proposed OANFIS WT method is more effective in reducing random noise and preserving relevant signal information than other contrastive methods.