Impact feature recognition method for non-stationary signals based on variational modal decomposition noise reduction and support vector machine optimized by whale optimization algorithm

Abstract

It is difficult to effectively distinguish the key information of non-stationary dynamic signals in many engineering applications, such as fault detection, geological exploration, and logistics transportation. To deal with this problem, a classification and recognition algorithm based on variational mode decomposition (VMD) and the Support Vector Machine (SVM) optimized by the Whale Optimization Algorithm (WOA) optimization model is first proposed in this study. The algorithm first applies VMD to decompose the non-stationary time-domain signals into multiple variational intrinsic mode functions (VIMFs). Then, it calculates the correlation coefficient between each mode and the original signals and conducts signal reconstruction by sorting the VIMFs. On the base of this, it performs modal filtering on the non-stationary signals according to the correlation coefficients between the reconstructed signal and the original signal. Subsequently, the WOA is used to optimize two key parameters of the SVM. Finally, the optimization model is exploited to classify and recognize the impact and vibration of non-stationary signals. A series of simulations and experiments for the algorithm is carried out and analyzed deeply. The comparative test results indicate that the classification and recognition method for non-stationary signals based on VMD and WOA-SVM (VMD-WOA-SVM) proposed in this paper converges faster and recognizes the key information of non-stationary dynamic signals more accurately with a recognition precision of 96.66%.

High accuracy key feature extraction approach for the non-stationary signals measurement based on NGO-VMD noise reduction and CNN-LSTM

“…As non-stationary signals, vibration signals are prone to being affected by excessive random noise during feature extraction, resulting in the submersion of key feature information. Meanwhile, the majority of noise is focused within the high-frequency range, and the frequency-domain characteristics of impact are predominantly distributed in lower sections, relatively speaking [30]. Considering the good filtering properties and sparsity of VMD, we propose the NGO-VMD filtering noise reduction method, which combines linear correlation to filter and reconstruct vibration signals.…”

Application of Variational Mode Decomposition and Whale Optimization Algorithm to Laser Ultrasonic Signal Denoising

“…Before this maximum, the noise is the main part of the IMFs, and the rest is the main part of the signal. Xu et al [ 19 ] calculated the correlation with the original function sequentially by gradually increasing the accumulated mode components and selected the mode component accumulated when the threshold is reached for the first time as the denoised signal. Although this method can obtain reasonable results, the threshold of the approximate entropy needs to be preset.…”