Way For Fault Diagnosis Research Articles

Multi-source domain adaptation using diffusion denoising for bearing fault diagnosis under variable working conditions

Transfer learning of multi-source domain adaptation seems a promising way for fault diagnosis of roller element bearings under variable working conditions. Data imbalance affects the performance of multi-source domain adaptation greatly and is expected to be solved by GAN. However, GAN-based transfer learning diagnosis models suffer pattern collapse and training instability, leading to unsatisfying diagnosis results in practical engineering. This paper proposes a denoising diffusion multi-source domain adaptation model (DDMDA). The proposed model uses diffusion denoising, which has better performance and is simpler to train than GAN, to generate shifted source domains for solving the data imbalance problem. A new noise prediction structure in diffusion denoising named Utrans-net, is constructed to restore the data distribution in the shifted source domain. Also, a multiple-domain discriminator structure is designed to extract features from multiple source domains to solve the issue of variable working conditions. Advanced models are used in this paper to compare with the proposed model for validation. Experimental demonstrations show that the proposed model is superior to the comparison models with satisfying performance.

From Complexity to Clarity: M2KCSVA's Nonlinear Temporal Correlation Analysis and Stationary Estimation Pave the Way for Fault Diagnosis in Ironmaking Processes

From Complexity to Clarity: M2KCSVA's Nonlinear Temporal Correlation Analysis and Stationary Estimation Pave the Way for Fault Diagnosis in Ironmaking Processes

Self-matching extraction fractional wavelet transform for mechanical equipment fault diagnosis

Time-frequency analysis (TFA) of vibration signals is an important way for fault diagnosis and predictive maintenance of mechanical equipment. However, the existing TFA methods, such as wavelet transform (WT) and synchrosqueezing transform, are incapable to describe local details of strongly time-varying signals, bringing such problems as time-frequency energy dispersion. Theoretically, it is more beneficial for time-varying band-limited signals to make multi-resolution analysis in the time-fraction frequency domain because such action can increase the selection chances of mother wavelets in any direction on the time-frequency plane by changing the angle . In order to improve the ability of analyzing strongly time-varying signals and the readability of the time-frequency plane, this paper has proposed a novel TFA method called self-matching extraction fractional WT (SMEFRWT). Firstly, within the theoretical framework of fractional WT (FRWT), a minimum entropy-based method is proposed to choose an optimal angle in the fractional domain. Such action can obtain the best aggregation angle of time-frequency energy and facilitate the separation of multi-component signals. Then, self-matching extraction operator, one instantaneous-frequency estimated operator that match the time-frequency structure of amplitude modulation signals, is constructed. The operator allows for the distribution of time-frequency energy of fast-changing signals in both frequency and time directions, making it possible to match the time-varying vibration signals from mechanical equipment more accurately. Finally, SMEFRWT is adopted to make TFA and extract fault features of the faulty experiment benches on gear and bearings, with its feasibility validated by experimental results.

Fault Diagnosis of Rolling Bearing Based on a Priority Elimination Method.

Aiming at the fault diagnosis accuracy of rolling bearings is not high enough, and unknown faults cannot be correctly identified. A priority elimination (PE) method is proposed in this paper. First, the priority diagnosis sequence of faults was determined by comparing the ratios of the inter-class distance to the intra-class distance for all faults. Then, the model training and fault diagnosis were carried out in order of the priority sequence, and the samples of the fault that had been identified were eliminated from the data set until all faults were diagnosed. For the diagnosis model, the stacked sparse auto-encoder network (SSAE) was selected to extract the features of the vibration signal. The extreme gradient boosting algorithm (XGBoost) was chosen to identify the fault type. Finally, the method was tested and verified by experimental data and compared with classical algorithms. Research results indicate the following: (1) with the addition of PE based on SSAE-XGBoost, the fault diagnosis accuracy can be improved from 96.3% to 99.27%, which is higher than other methods; (2) for the test set with the samples of unknown faults, the diagnosis accuracy of SSAE-XGBoost with PE can reach 92.34%, which is nearly 6% higher than that without PE and is also obviously higher than other classical fault diagnosis methods with or without PE. The PE method can not only improve the diagnosis accuracy of faults but also identify unknown faults, which provides a new method and way for fault diagnosis.

Benign-to-malignant transition in external short circuiting of lithium-ion batteries

External short circuiting (ESC) is a main source of battery faults. However, the ESC damage mechanism and its evolution process are unclear, resulting in difficulties in safety management. Here, we report the impact of different ESC durations on battery performance and divide the ESC process into four stages. The analysis reveals that there is a benign regime in which the ESC effect is comparable to nominal charge-discharge cycling. When ESC duration is close to a critical time, there is a benign-to-malignant transition. The critical time corresponds to the inflection point at the end of battery current curve, which reflects heating-induced membrane melting and contraction that distorts the jellyroll. The critical time surface under various conditions is established and the center temperatures of the critical time and battery completely failure are disclosed, paving the way for fault diagnosis and using “benign ESC” in a controlled fashion. • Benign-to-malignant evolution of battery ESC damage mechanism is revealed • Critical time to characterize the occurrence of malignant battery damage is identified • Equivalent influence correlation is established between benign SC and nominal cycling • Battery heating method based on controllable benign short circuit is proposed Chen et al. reveal the evolution of damage mechanism during battery external short circuit, pointing out that there is a benign-to-malignant transition. The critical time to characterize the battery malignant damage is identified. This research may open new possibilities for applying short circuit in a controlled fashion.

Trackside acoustic detection of axle bearing fault using wavelet domain moving beamforming method

Trackside acoustic detection is a promising way for fault diagnosis of train axle bearings. There are two main tough issues in the trackside acoustic detection, which are the signal distortion due to the Doppler effect and strong noise interference from other subsystems of the vehicles. This study presents a solution to overcome both issues for axle bearing fault diagnosis by using a Wavelet domain Moving Beamforming (WMB) method. Firstly, the time domain acoustic signal acquired by a microphone array is transformed to the wavelet domain, and the Doppler effect is removed through a proposed wavelet domain resampling method. Then, the wavelet domain beamforming is applied to enhance the signal of axle bearing and reduce the interference noise. Afterwards, according to the kurtoses of frequency components, the components with smaller kurtoses are discarded to further enhance the impulsive component carrying the fault information. The train axle bearing faults can finally be diagnosed by observing the envelope spectrum of the processed signal. The results of numerical simulations and experiments verify the feasibility of the WMB. The influence of measurement setup parameters is also investigated and discussed.

Spatio-temporal fusion neural network for multi-class fault diagnosis of wind turbines based on SCADA data

Abstract Numerous sensors have been deployed in different locations in components of wind turbines to continuously monitor the health status of the turbine system and accordingly, generate a large volume of operation data by the supervisory control and data acquisition (SCADA) system. Naturally, these sensory data are multivariate time series with high spatio-temporal correlations. It is still challenging to effectively model such correlations and then enable an accurate fault diagnosis. To this end, we proposed a new spatio-temporal fusion neural network (STFNN) for wind turbine fault diagnosis. Specifically, a multi-kernel fusion convolution neural network (MKFCNN) with multiple convolution kernels of different sizes is first designed to extract multi-scale spatial correlations among different variables. Then, we adopt the long short-term memory (LSTM) to further learn the temporal dependence of the learned spatial features. The proposed STFNN model provides an end-to-end fault diagnosis way, which can directly learn spatio-temporal dependency from the raw SCADA data and give the fault diagnosis result. The effectiveness and superiority of the proposed method are evaluated on a generic wind turbine benchmark simulation dataset and a SCADA dataset from a real wind farm. Both experimental results have indicated that the proposed method outperformed several compared methods.

Time—window Complexity and its Application in the Fault Diagnosis of Bearing

Different from conventional spectral method, complexity analysis treaded with the signals’ time domain structure feature. Before, due to the no stationary and the unevenness of state space of mechanical signals, the state information is likely to get a loss. Now, the time-window complexity is proposed to overcome certain limitations of complexity itself in some extent. It will help to extract the state features of mechanical systems in different states. The concept and algorithm of time-window complexity is introduced detailed. The way of applying the time-window complexity for fault diagnosis is discussed. The mechanical signals of ball bearing with slight flaw and that of four kinds of impact-rubbing states of a typical rotor are then studied. The results show that time-window complexity can reflect the early fault of ball bearing, and differentiate the four kinds of rotor impact-rubbing states properly, which provides another effective way for fault diagnosis of mechanical systems.

Fault Diagnosis Expert System for Special Electronic Equipment Based on the Fuzzy Neural Network

In order to improve the efficiency and reliability of fault diagnosis for the special electronic equipment, an intelligent fault diagnostic model based on Fuzzy Neural Network (FNN) is proposed. Firstly, the fault diagnosis model based on the FNN Expert System (ES) is built. Secondly, the fault diagnosis expert system of the special electronic equipment based on this model is introduced. Finally, experiments show that the proposed model is correct and the FD system is effective. Moreover, the given method provides a new way of fault diagnosis for other modern electronic system.

Transformer Fault Dignosis Based on Feature Selection and Parameter Optimization

Failure of transformer is very complex, dissolved Gas in Oil Analysis (DGA) is presently the easier and simpler way for fault diagnosis of oil-immersed transformers.The correct selection of features of dissolved gas data can improve efficiency of transformer fault diagnosis. SVM is more effective than traditional methematic model to discribe the type of fault of transformer.As for the problem of difficulty of determining parameters in SVM applications, genetic algorithm (GA) was used to select SVM parameters. The test results show that this GA-SVM model is effective to detect failure of transformer.

Inner Leakage Fault Diagnosis of Hydraulic Cylinder Using Wavelet Energy

Inner leakage of hydraulic cylinder is a very serious failure in the hydraulic system and it can lead to many problems.A important fault diagnosis way is to detect the pressure signal.But the pressure signal is seriously influenced by pressure fluctuation and other noises.It is difficult to extract features from pressure singal. Aiming at the difficulty in extracting feature from pressure singal in fault diagnosis for leakage of hydraulic cylinder,a fault diagnosis approch based on monitoring preesure singal and wavelet energy is proposed in this article.According to the method, the enegry of different frequency bands after wavelet decomposition costitutes the eigenvectors at first, then these eigenvectors are input into BP network to identify faults. The experimental results show that three modes of no leakage, slighter leakage and severe leakage were correctly identified.This approach can be used in the leakage fault diagnosis of hydraulic cylinder.

Design of heartbeat invalidation detecting mechanism in triple modular redundancy multi‐machine system

PurposeAccording to the high‐reliability requirement and especial structure of triple modular redundancy (TMR) emergent trip multi‐machine system, the purpose of this paper is to introduce a highly effective invalidation detecting mechanism based on heartbeat mechanism.Design/methodology/approachThe structure of TMR system is designed to be multi‐master and multi‐save. According to the multi‐machine environment, an invalidation detecting mechanism is designed. Considering the advantage and weakness of the traditional PULL and PUSH heartbeat model, the mechanism combines PUSH model with PULL model. To further enhance real‐time ability and reliability of the invalidation detecting mechanism, hierarchy heartbeat monitor and integrating the heartbeat message into data transmission are designed in the mechanism.FindingsThe invalidation detecting mechanism combines PUSH with PULL heartbeat monitor model can enhance reliability and decrease the system misdiagnosis rate greatly. By adapting hierarchy heartbeat mechanism and integrating the heartbeat detecting message into the data transmission, the system communication cost is decreased and the input‐to‐output response time is shortened. The effectiveness and reliability of this mechanism are proved by system field commissioning and MATLAB simulation.Originality/valueUsing a new heartbeat invalidation detecting mechanism, this paper provides a new way of fault diagnosis of TMR multi‐machine system.

On improving training time of neural networks in mixed signal circuit fault diagnosis applications

Testing issues are becoming more and more important with the quick development of both digital and analog circuit industry. Analog-to-digital converters (ADCs) are becoming more and more widespread owing to their fundamental capacity of interfacing analog physical world to digital processing systems. In this paper, we study the use of neural networks in fault diagnosis of ADCs and compare the results with other ADC testing approaches such as histogram, FFT and sinewave curve fit test techniques. In this paper, we introduced the idea of separation of neural network’s output matrix to improve the training phase time, called ‘index-separation’ approach. Finally, we concluded that training time in this method is about 0.25 times as much as that in the normal training method. We also concluded that this approach does not affect network’s decision strength. Besides, we concluded that if the complexity of the circuit increases, this method will still be effective. Therefore, this method is a robust way for fault diagnosis of mixed signal circuits.

Online Rotor Mixed Fault Diagnosis Way Based on Spectrum Analysis of Instantaneous Power in Squirrel Cage Induction Motors

Broken rotor bars and eccentricity are common faults in squirrel cage induction motors. These two faults usually occur simultaneously. This paper will deal with this kind of mixed fault. It is well known that the characteristic frequency of broken bars in the stator line current of a squirrel cage induction motor is very near to that of the fundamental component. In the spectrum of the stator line current, the characteristic components related to broken rotor bars are always submerged by the fundamental one, and it is difficult to detect the broken bar fault at an early stage. In our work, instantaneous power of the motor is used as the quantity to be monitored. Theoretical analysis indicates that the spectrum of ac level of the spectrum of the instantaneous power is clear from any component at the fundamental supply frequency, and the fault characteristics can be highlighted, which is effective toward the separation of mixed faults and the quantification of the fault extent. Experimental results have demonstrated the effectiveness of the proposed technique.

Fault Diagnosis in HVDC Systems with Neural Networks

This paper describes a neural network design and its simulation results for fault diagnosis in HVDC power systems. Fault diagnosis is carried out by mapping input data patterns, which represent the behaviour of the system, to one or more fault conditions. The behaviour of the converters is described in terms of the time varying patterns of conducting thyristors, pulse zone periods, voltage zone periods and ac & dc fault characteristics. A three-layer neural network consisting of 24 input nodes, 12 hidden nodes and 13 output nodes are used. 13 different faults were considered and the neural network approach shows a great potential as an effective way for fault diagnosis.

Identification of Fuzzy Rule Based System for Fault Diagnosis in Chemical Plants

A simplified fuzzy reasoning algorithmic approach for describing the cause -effect relationships in a real technological system is proposed. It uses rule based fuzzy model with simple triangular shape of all the membership functions. In order to determine the number of the fuzzy sets (membership functions) as well as the number of the fuzzy rules, an identification algorithm is proposed. The adjustment of the right part of the rules is done by using real input-output data and Steepest Descent Method. The main features of the proposed identification method are discussed on a numerical example for fuzzy modelling. A method for use of the fuzzy models created in this way for fault diagnosis of an industrial system is also proposed and discus sed in the paper.