Fig 2 - available via license: CC BY
Content may be subject to copyright.
Source publication
To achieve the goal of automated rolling bearing fault diagnosis, a variational mode decomposition (VMD) based diagnosis scheme was proposed. VMD was firstly used to decompose the vibration signals into a series of band-limited intrinsic mode functions (BLIMFs). Subsequently, the multiscale fractal dimension (MSFD) and multiscale energy (MSEN) of e...
Context in source publication
Context 1
... the subset, the obtained number of samples was 116. Among them, 58 samples were randomly selected as training set and the remaining samples were used as testing set. A detailed description of the segmented vibration signals is shown in Table 1 and the time domain waveforms of these seven classes of rolling bearing vibration signals are shown in Fig. 2. ...
Similar publications
Bearings are critical parts of rotating machines, making bearing fault diagnosis based on signals a research hotspot through the ages. In real application scenarios, bearing signals are normally non-linear and unstable, and thus difficult to analyze in the time or frequency domain only. Meanwhile, fault feature vectors extracted conventionally with...
Citations
... To diagnose bearing faults [20], constructed an integrated weighted arithmetic mean-convolutional neural network model. Based on prototype learning network, adversarial semi-supervision (DASS) has been proposed in [21], which was used for fault diagnosis of rolling bearing vibration signals [22]. Used variational mode decomposition to decompose vibration signals and used genetic algorithms to optimize support vector machine parameters to give an improved classification performance [23]. ...
Motor vibration signal data sets are characteristically random and nonlinear, and its features are difficult to extract for fault identification. To reduce the uncertainty of fault diagnosis, a method based on principal component analysis (PCA) and Discrete Belief Rule Base (DBRB) was developed for the first time. Initially, the vibration signal was first denoised using a wavelet threshold algorithm to eliminate interference. Second, overlapping signals were segmented into 15 time windows and a total of 13 typical time domain features and mathematical statistical features were extracted. Third, the dimensions of the features were reduced to three principal components by principal component analysis (PCA) and were taken as the antecedent attributes of the DBRB. However, the amount of information in each principal component is different, so the variance contribution rate was taken as an antecedent attribute weight to restore the original data characteristics. Fourth, a PCA-DBRB model was established, which effectively avoided the combinatorial explosion problem of rule base in the DBRB model. In addition, to obtain appropriate reference values, the k-means algorithm was introduced to take the cluster centers as reference values. The method was then validated by collecting typical fault data from motor bench experiments. The results demonstrated that compared with other traditional classifiers, this approach is more effective and superior in classification performance and more accurate in diagnosing faults from motor vibration data.
... For bearing fault diagnosis, approximate entropy [32], sample entropy [10,32,33], Renyi entropy [34,35], permutation entropy [13,22,26,28], and fuzzy entropy [20,36] have been explored [13,37]. Few features such as fractal dimension [38], higher-order spectra entropy [39], Tsallis entropy [40], modified multiscale entropy, and log energy have not been explored for bearing fault detection under varying speed, despite their superior performance to process nonlinear and non-stationary signals [41]. Therefore, this study proposed to extract various entropies for all the decomposed MFs, including those which have not been explored for non-stationary signals of mechanical systems. ...
... Over the last decade, researchers have extensively used classifiers to compare and improve the performance of automated methods for fault diagnosis. To name a few, neural network [2,5,13,25,37], support vector machine (SVM) [20,22,25,31,38,40], k-nearest neighbor [14,22,36], extreme learning machine [22,26], random forest (RF) [13,36], decision tree [43], etc. have been used to classify the bearing faults. These were employed at constant speed conditions, and their performances at time-varying speed are yet to be explored. ...
... FD can characterize nonlinear information of a signal of rotating machinery qualitatively and quantitatively [38]. FD quantifies the complexity and the self-similarity of an object. ...
This work proposes a systematic approach to detect and classify bearing faults using vibration signals under varying speeds. The proposed approach consists of several steps, such as segmentation of signal consisting of maximum fault relevant information and extraction of features less influenced by varying speeds, and develop a machine learning model for online classification of bearing faults. Bearing when operating under time-varying speeds, the most critical and challenging step, is the demodulation of non-stationary and nonlinear vibration signals exhibiting severe modulations. The empirical wavelet transformation (EWT) algorithm has been used to decompose the raw signal into multiple mode functions (MFs), thereby detecting faults. However, these MFs contaminated by vestigial noise, when processed, mislead the detection of incipient bearing faults, thereby reducing EWT performance. Hence, this study addresses this by proposing the selection of the most impulsive MF for varying speed by estimating instantaneous frequency, which lies near bearing characteristic defect frequencies, thereby eliminating the possibility of vestigial noise being processed. Further, ten entropies, root-mean-square, and kurtosis are computed from the selected MF for statistical analysis. The results of the proposed approach are compared with the ensemble empirical mode decomposition to highlight the capabilities. Statistically significant fault discriminating features are identified using the Kruskal–Wallis test. These identified features are subsequently utilized by the Random Forest classifier. Thus, it has resulted in higher accuracy in detecting and classifying the different faults trapped by severe modulations.
... The IMF in EEMD is characterized as the mean of an ensemble of trials whereby a finite-amplitude white noise signal is added to the decomposed data in each trial; this approach increases the computational burden since the data size of IMF is equal to that of the raw data. In recent years, VMD has been introduced for noise analyses of rotating machines and as a fault diagnosis method that has shown very promising results [6,20,21]. Although signals are separated into a series of IMFs, IMFs depend on the values of the balancing parameter and the number of modes that are adjustable, and the results may be inaccurate when they are not set in place [22]. ...
... and crossover percentage of the population=0.8. The input ranges of a and K are set to [100, 10000] and [2,20] respectively. After 50 runs of GA, the average values are a = 1463 and K = 9.9 respectively. ...
Automotive engine knock is an abnormal combustion phenomenon that affects engine performance and lifetime expectancy, but it is difficult to detect. Collecting engine vibration signals from an engine cylinder block is an effective way to detect engine knock. This paper proposes an intelligent engine knock detection system based on engine vibration signals. First, filtered signals are obtained by utilizing variational mode decomposition (VMD), which decomposes the original time domain signals into a series of intrinsic mode functions (IMFs). Moreover, the values of the balancing parameter and the number of IMF modes are optimized using genetic algorithm (GA). IMFs with sample entropy higher than the mean are then selected as sensitive subcomponents for signal reconstruction and subsequently removed. A multiple feature learning approach that considers time domain statistical analysis (TDSA), multi-fractal detrended fluctuation analysis (MFDFA) and alpha stable distribution (ASD) simultaneously, is utilized to extract features from the denoised signals. Finally, the extracted features are trained by sparse Bayesian extreme learning machine (SBELM) to overcome the sensitivity of hyperparameters in conventional machine learning algorithms. A test rig is designed to collect the raw engine data. Compared with other technology combinations, the optimal scheme from signal processing to feature classification is obtained, and the classification accuracy of the proposed integrated engine knock detection method can achieve 98.27%. We successfully propose and test a universal intelligence solution for the detection task.
... It is widely used in fault diagnosis [16]. Chen et al. proposed the fault diagnosis of rolling bearing based on VMD and support vector machine (SVM), which significantly improved fault identification accuracy through multiscale fractal dimension and multiscale energy calculation features [17]. ...
Rolling bearing fault diagnosis is a meaningful and challenging task. Most methods first extract statistical features and then carry out fault diagnosis. At present, the technology of intelligent identification of bearing mostly relies on deep neural network, which has high requirements for computer equipment and great effort in hyperparameter tuning. To address these issues, a rolling bearing fault diagnosis method based on the improved deep forest algorithm is proposed. Firstly, the fault feature information of rolling bearing is extracted through multigrained scanning, and then the fault diagnosis is carried out by cascade forest. Considering the fitting quality and diversity of the classifier, the classifier and the cascade strategy are updated. In order to verify the effectiveness of the proposed method, a comparison is made with the traditional machine learning method. The results suggest that the proposed method can identify different types of faults more accurately and robustly. At the same time, it has very few hyperparameters and very low requirements on computer hardware.
In order to improve the accuracy of short-term photovoltaic power prediction, a comprehensively artificial intelligent (AI) based algorithm has been proposed. Firstly, a fuzzy C-means (FCM) is used to cluster historical data into different groups according to meteorological variables; Secondly, a variational mode decomposition (VMD) is used to divide original signals into multi-frequency components as a series of intrinsic mode functions (IMF) while whale optimization algorithm (WOA) externally optimizes the parameters of VMD so as to stimulate the data processing. Least squares support vector machine (LSSVM) is adapted to build up the prediction framework with an assistance from improved sparrow search algorithm (ISSA) that aims to solve the key issues on local optimization over LSSVM training process. Under the proposed framework, practical power signals from Northwest China has been tested. The outcomes verify the effectiveness of this comprehensive algorithm. The indicator, mean absolute percentage error (MAPE), has decreased by 17.17%, 15.78% and 5.37% while R2 has increased by 0.17, 0.15 and 0.03 comparing with LSSVM, ISSA-LSSVM and VMD-LSSVM. Moreover, the study outcomes have also evidenced the superiority of ISSA over other fellow heuristic algorithms such as PSO, QPSO and SSA in both convergence speed and accuracy when an ISSA facilitates LSSVM for prediction.
Many works of bearing fault diagnosis based on vibration signals have been present. However, most of them work under ideal conditions that the fault data are enough, but it is unrealistic in real applications. In this paper, we proposed a method to address the problem. It is a pipeline method of ensemble feature engineering, data improvement, and classifier design. First, multiple features of the bearing vibration signal are extracted. And then, the fault data are augmented by a novel oversampling method. Finally, the faults with a stacking classifier were identified. The proposed oversampling method can remove noise instances before and after oversampling, to overcome the risk of synthesizing noisy instances. It is not a new oversampling paradigm, but an enhancement module of the existing methods. Relying on it, four common oversampling methods improve the classification performance by 1–2% on publicly available NASA software defect datasets. Empirical results indicate the superiority of the proposed method: When the imbalanced ratio is 10:1, our method can obtain 99.2% and 91.5% accuracy rates on CWRU and Paderborn datasets, respectively.
