RESUMO
The application of transfer learning in fault diagnosis has been developed in recent years. It can use existing data to solve the problem of fault recognition under different working conditions. Due to the complexity of the equipment and the openness of the working environment in industrial production, the status of the equipment is changeable, and the collected signals can have new fault classes. Therefore, the open set recognition ability of the transfer learning method is an urgent research direction. The existing transfer learning model can have a severe negative transfer problem when solving the open set problem, resulting in the aliasing of samples in the feature space and the inability to separate the unknown classes. To solve this problem, we propose a Weighted Domain Adaptation with Double Classifiers (WDADC) method. Specifically, WDADC designs the weighting module based on Jensen-Shannon divergence, which can evaluate the similarity between each sample in the target domain and each class in the source domain. Based on this similarity, a weighted loss is constructed to promote the positive transfer between shared classes in the two domains to realize the recognition of shared classes and the separation of unknown classes. In addition, the structure of double classifiers in WDADC can mitigate the overfitting of the model by maximizing the discrepancy, which helps extract the domain-invariant and class-separable features of the samples when the discrepancy between the two domains is large. The model's performance is verified in several fault datasets of rotating machinery. The results show that the method is effective in open set fault diagnosis and superior to the common domain adaptation methods.
RESUMO
Though accelerometers for condition diagnosis of a bearing is preferably placed at the nearest position of the bearing as possible, in some plant equipment, the accelerometer is difficult to set near the diagnosed bearing, and in many cases, sensors have to be placed at a location far from the diagnosed bearing to measure signals for diagnosing bearing faults. Since, in these cases, the measured signals contain stronger noise than the signal measured near the diagnosed bearing, bearing faults are more difficultly to be detected. In order to overcome the above difficulty, this paper proposes a new fault auto-detection method by which the signals measured by an accelerometer located at a far point from the diagnosed bearing can be used to simply and accurately detect the bearing faults automatically. Firstly, the hybrid GA (the combination of genetic algorithm and tabu search) is used to automatically search and determine the optimum cutoff frequency of the high-pass filter to extract the fault signal of the abnormal bearing. Secondly, the bearing faults are precisely diagnosed by possibility theory and fuzzy inference. Finally, in order to demonstrate the effectiveness of these proposed methods, these methods were applied to bearing diagnostics using vibration signals measured at the far point of the diagnostic bearing, and the efficiency of these methods was verified by the results of automatic bearing fault diagnosis.
RESUMO
In order to separate and extract compound fault features of a vibration signal from a single channel, a novel signal separation method is proposed based on improved sparse non-negative matrix factorization (SNMF). In view of the traditional SNMF failure to perform well in the underdetermined blind source separation, a constraint reference vector is introduced in the SNMF algorithm, which can be generated by the pulse method. The square wave sequences are constructed as the constraint reference vector. The output separated signal is constrained by the vector, and the vector will update according to the feedback of the separated signal. The redundancy of the mixture signal will be reduced during the constantly updating of the vector. The time-frequency distribution is firstly applied to capture the local fault features of the vibration signal. Then the high dimension feature matrix of time-frequency distribution is factorized to select local fault features with the improved SNMF method. Meanwhile, the compound fault features can be separated and extracted automatically by using the sparse property of the improved SNMF method. Finally, envelope analysis is used to identify the feature of the output separated signal and realize compound faults diagnosis. The simulation and test results show that the proposed method can effectively solve the separation of compound faults for rotating machinery, which can reduce the dimension and improve the efficiency of algorithm. It is also confirmed that the feature extraction and separation capability of proposed method is superior to the traditional SNMF algorithm.
RESUMO
Singular value decomposition (SVD) is an effective method used in bearing fault diagnosis. Ideally two important problems should be solved in any diagnosis: one is how to decide the dimension embedding of the trajectory matrix (TM); the other is how to select the singular value (SV) representing the intrinsic information of the bearing condition. In order to solve such problems, this study proposed an effective method to find the optimal TM and SV and perform fault signal filtering based on false nearest neighbors (FNN) and statistical information criteria. First of all, the embedded dimension of the trajectory matrix is determined with the FNN according to the chaos theory. Then the trajectory matrix is subjected to SVD, which is helpful to acquire all the combinations of SV and decomposed signals. According to the similarities of the signal changed back and signal in normal state based on statistical information criteria, the SV representing fault signal can be obtained. The spectrum envelope demodulation method can be used to perform effective analysis on the fault. The effectiveness of the proposed method is verified with simulation signals and low-speed bearing fault signals, and compared with the published SVD-based method and Fast Kurtogram diagnosis method.
RESUMO
Fault transient impulses induced by faulty components in rotating machinery usually contain substantial interference. Fault features are comparatively weak in the initial fault stage, which renders fault diagnosis more difficult. In this case, a sparse representation method based on the Majorzation-Minimization (MM) algorithm is proposed to enhance weak fault features and extract the features from strong background noise. However, the traditional MM algorithm suffers from two issues, which are the choice of sparse basis and complicated calculations. To address these challenges, a modified MM algorithm is proposed in which a sparse optimization objective function is designed firstly. Inspired by the Basis Pursuit (BP) model, the optimization function integrates an impulsive feature-preserving factor and a penalty function factor. Second, a modified Majorization iterative method is applied to address the convex optimization problem of the designed function. A series of sparse coefficients can be achieved through iterating, which only contain transient components. It is noteworthy that there is no need to select the sparse basis in the proposed iterative method because it is fixed as a unit matrix. Then the reconstruction step is omitted, which can significantly increase detection efficiency. Eventually, envelope analysis of the sparse coefficients is performed to extract weak fault features. Simulated and experimental signals including bearings and gearboxes are employed to validate the effectiveness of the proposed method. In addition, comparisons are made to prove that the proposed method outperforms the traditional MM algorithm in terms of detection results and efficiency.
RESUMO
Compound faults often occur in rotating machinery, which increases the difficulty of fault diagnosis. In this case, blind source separation, which usually includes independent component analysis (ICA) and sparse component analysis (SCA), was proposed to separate mixed signals. SCA, which is based on the sparsity of target signals, was developed to sever the compound faults and effectively diagnose the fault due to its advantage over ICA in underdetermined conditions. However, there is an issue regarding the vibration signals, which are inadequately sparse, and it is difficult to represent them in a sparse way. Accordingly, to overcome the above-mentioned problem, a sparsity-promoted approach named wavelet modulus maxima is applied to obtain the sparse observation signal. Then, the potential function is utilized to estimate the number of source signals and the mixed matrix based on the sparse signal. Finally, the separation of the source signals can be achieved according to the shortest path method. To validate the effectiveness of the proposed method, the simulated signals and vibration signals measured from faulty roller bearings are used. The faults that occur in a roller bearing are the outer-race flaw, the inner-race flaw and the rolling element flaw. The results show that the fault features acquired using the proposed approach are evidently close to the theoretical values. For instance, the inner-race feature frequency 101.3 Hz is very similar to the theoretical calculation 101 Hz. Therefore, it is effective to achieve the separation of compound faults utilizing the suggest method, even in underdetermined cases. In addition, a comparison is applied to prove that the proposed method outperforms the traditional SCA method when the vibration signals are inadequate.
RESUMO
The traditional approaches for condition monitoring of roller bearings are almost always achieved under Shannon sampling theorem conditions, leading to a big-data problem. The compressed sensing (CS) theory provides a new solution to the big-data problem. However, the vibration signals are insufficiently sparse and it is difficult to achieve sparsity using the conventional techniques, which impedes the application of CS theory. Therefore, it is of great significance to promote the sparsity when applying the CS theory to fault diagnosis of roller bearings. To increase the sparsity of vibration signals, a sparsity-promoted method called the tunable Q-factor wavelet transform based on decomposing the analyzed signals into transient impact components and high oscillation components is utilized in this work. The former become sparser than the raw signals with noise eliminated, whereas the latter include noise. Thus, the decomposed transient impact components replace the original signals for analysis. The CS theory is applied to extract the fault features without complete reconstruction, which means that the reconstruction can be completed when the components with interested frequencies are detected and the fault diagnosis can be achieved during the reconstruction procedure. The application cases prove that the CS theory assisted by the tunable Q-factor wavelet transform can successfully extract the fault features from the compressed samples.
RESUMO
This paper proposes an intelligent sequential diagnosis method for plant machinery using statistical filter (SF), signal histogram and genetic programming (GP). The SF is used to cancel noise from the measured vibration signal for raising the accuracy of fault diagnosis. Since the vibration signal measured for the condition diagnosis conforms to various probability distributions, histograms are used to reflect the signal features instead of the conventional symptom parameters (SPs). Then, the genetic programming (GP) is used to generate new variables termed "integrated symptom parameters" (GP-ISPs) from the histogram. GP-ISPs obtained by the auto-reorganized histogram can reflect features and raise the sensitivity of the fault diagnosis by the greatest amount possible. Furthermore, a sequential diagnosis algorithm using GP-ISPs is also proposed to realize precise diagnosis for distinguishing fault types. Finally, the effectiveness of the proposed method is verified by applying it to the fault diagnosis of a centrifugal blower. The proposed method has wide applicability and is practical in the field of machinery fault diagnosis.