A rotating machinery fault feature extraction approach based on an adaptive wavelet denoising method and synthetic detection index

Self Cite

Rotating machinery feature extraction is critical for the subsequent fault diagnosis and ensuring safe and stable operation. However, the commonly used methods often have limitations, such as the extracted features being redundant or insufficient and the method parameters generally being set based on experience. Although some feature extraction methods apply optimization algorithms for parameter setting, their objective functions are often too simple to obtain favorable fault diagnosis results. To address these issues, a target detection index (TDI) is constructed which can consider the discrimination among features more comprehensively and make the extracted features more sensitive. Furthermore, a fault feature extraction method is proposed based on TDI and successive variational mode decomposition (SVMD). Taking TDI as the objective function, genetic algorithm (GA) is used for the feature selection process and SVMD parameter optimization. The obtained features are then fused and visualized using t-distributed stochastic neighbor embedding and are classified using support vector machines. The Case Western Reserve University data and hydropower generating unit data are employed for method verification. When compared with other decomposition algorithms, the proposed method exhibits great ability at extracting highly sensitive features.

Section: Feature Parametersmentioning

confidence: 99%

A feature extraction method for rotating machinery fault diagnosis based on a target detection index and successive variational mode decomposition

Cao,

Zhang,

et al. 2023

Self Cite

“…In contemporary research, prevalent signal preprocessing techniques encompass the wavelet packet transform (WPT), empirical mode decomposition (EMD) and ensemble EMD (EEMD) [15]. Zhou et al proposed an adaptive wavelet denoising method to decompose the vibration signals of rotating machinery [16]. Shrivastava et al used EMD and EEMD to extract the chatter frequency of the CNC tool during machining to monitor its health condition [17].…”

Section: Introductionmentioning

confidence: 99%

Optimised LightGBM-based health condition evaluation method for the functional components in CNC machine tools under strong noise background

Jia,

Jialong,

Wanghao

et al. 2024

The accurate health condition evaluation of the functional components in computer numerical control (CNC) machine tools is an important prerequisite for predictive maintenance and fault warning. The vibration signals of the functional components in CNC machine tools often contain substantial noise, impeding the extraction of relevant health condition information from the vibration signals. This work presents an approach that leverages the variational mode decomposition (VMD) enhanced by the Artificial Hummingbird Algorithm (AHA) alongside the Light Gradient Boosting Machine (LightGBM) optimised through particle swarm optimisation (PSO) to evaluate the health condition of the functional components in CNC machine tools amidst pervasive noise. Initially, the AHA optimised the penalty factor (α) and the decomposition layer (K) within the VMD. This optimised VMD was subsequently applied to denoise the original vibration signals. After this denoising process, PSO was employed to optimise the learning rate and maximum tree depth within LightGBM. Health condition evaluation experiments were executed on the feed system and spindle of the CNC machine tool to validate the proposed methodology. Comparative analysis indicates that the proposed method attains paramount accuracy and computational efficiency, which are crucial for accurately evaluating the health condition of the functional components in CNC machine tools.

“…Li et al [18] selected 4-layer wavelet decomposition and Sym6 wavelet basis function to denoise the AE signal of deep hole drilling, effectively removing the noise in the AE signal, and also preserving the integrity of the signal, avoiding signal distortion, which significantly improves the denoising effect. Zhou et al [19] proposed an adaptive wavelet denoising (AWD) method and evaluated the signal noise level using permutation entropy, proposing a new feature extraction method. The AWD method has better signal denoising performance for fault features.…”

Section: Introductionmentioning

confidence: 99%

Research on features and localization of AE signals from the mechanical body of industrial robots based on WD-EMD

Dong,

You,

et al. 2024

In order to reveal the propagation characteristics of acoustic emission (AE) signals in the body of industrial machinery, the characteristic frequencies and wave speeds of AE signals propagated on the interior and exterior surfaces (IS and ES) of the body were extracted. Subsequently, an algorithm utilizing characteristic frequency and time difference of arrival (TDOA) is proposed for the identification and localization of AE sources. Initially, an AE source is induced on the IS and ES of the body using the pencil-lead break method in accordance to ASTM standards, and the AE signal is captured by two piezoelectric sensors at a sampling frequency of 3 MHz. Then, to avoid the limitations of wavelet decomposition using self-selected wavelet scale and the problem that a single indicator cannot properly evaluate the correlation between independent mode functions (IMFs) with the original signal, this paper will conduct 4-layer wavelet decomposition of the original signal according to the response frequency range of the sensor, and select the wavelet details within the stable range of the response frequency of the sensor for preliminary reconstruction,and then the empirical mode decomposition (EMD) method is used to decompose the de-noised AE signal into 7 IMFs, and the AE waveform is reconstructed by the combined information of correlation coefficient and variance accounted for. Secondly, the reconstruction method combined with EMD analysis and a single index is compared with the proposed method in this paper to verify the reliability of the proposed method. In addition, the frequency domain characteristics of AE signal propagation process on the IS and ES of the body are extracted. Finally, based on the TDOA principle, the propagation speed of AE signal on the IS and ES of the body is calculated. Based on the geometric relationship between the AE source and two sensors, an algorithm for the location of the AE source is proposed. The results show that the proposed signal reconstruction method can effectively extract the features of AE signals, and the average positioning accuracy of the localization algorithm based on characteristic frequency and TDOA reaches 0.86%.