Adaptive Unsaturated Stochastic Resonance in Parallel Array and its Application to Acoustic Emission Signals

Among various signal processing approaches, stochastic resonance (SR) has been widely employed for weak signal detection and mechanical fault diagnosis. Various advancements have been focused on identifying useful information from the frequency domain by optimizing parameters in a post-processing environment to activate SR. Yet, these methods often require detailed information about the original signal a priori, which is challenging from measurements that are already overwhelmed by noise. Furthermore, classical bistable SR has often been employed for weak signal detection, which exhibits an inherent signal distortion due to output saturation that reduces the signal recovery performance. To address these concerns and advance the state of the art, we propose a novel signal denoising method that exploits unsaturated SR in a parallel array of piecewise continuous bistable systems. The original noise-contaminated signal is adaptively scaled by an optimal gain value that is determined from a non-dimensional model based on the attendant noise level, which is one of the few parameters that can be reliably identified from practical noise-contaminated signals. As a result, the proposed approach can operate without any post-processing optimization and parameter selection. Numerical investigations are performed with a simulated acoustic emission signal (amplitude modulated sine pulse) with various amplitudes and attendant noise levels to illustrate the operation principle and the effectiveness of the proposed approach. The results exemplify the promising potential of implementing the proposed approach for enhancing online signal denoising in practice.