Non-periodic Noisy Signals Denoising Using Adaptive Neuro-Fuzzy Inference System (ANFIS)

Abstract

Signal always occurs with noise. Since noise acts as an unwanted signal, we must clear or reduce it with some denoising method. It is relatively easy to denoise the normal distribution noise-contaminated the periodic signal. The problem ascends if a non-Gaussian noise intrudes into a non-periodic signal. The standard filter, such as DWT (discrete wavelet transforms), cannot overcome this directly and blindly. In this research, we proposed ANFIS (Adaptive Neuro-Fuzzy Inference System) as a non-periodic noisy signal denoising method. Foremost, the ANFIS trained to mimic or estimate the interfered noise, then this noise estimation used as a subtractive signal in a non-periodic noisy signal. As a result, the ANFIS can reduce the non-Gaussian noise in the various noisy non-periodic signals with minimum error better than standard DWT (Discrete Wavelet Transform).