•A feature-based transfer neural network is proposed for bearing fault diagnosis.•Diagnosis knowledge is transferred from laboratory bearings to locomotive bearings.•Multi-layer domain adaptation is ...used to correct discrepancy of learned features.•Pseudo label learning is used to reduce among-class distance of learned features.
Intelligent fault diagnosis of rolling element bearings has made some achievements based on the availability of massive labeled data. However, the available data from bearings used in real-case machines (BRMs) are insufficient to train a reliable intelligent diagnosis model. Fortunately, we can easily simulate various faults of bearings in a laboratory, and the data from bearings used in laboratory machines (BLMs) contain diagnosis knowledge related to the data from BRMs. Therefore, inspired by the idea of transfer learning, we propose a feature-based transfer neural network (FTNN) to identify the health states of BRMs with the help of the diagnosis knowledge from BLMs. In the proposed method, a convolutional neural network (CNN) is employed to extract transferable features of raw vibration data from BLMs and BRMs. Then, the regularization terms of multi-layer domain adaptation and pseudo label learning are developed to impose constraints on the parameters of CNN so as to reduce the distribution discrepancy and the among-class distance of the learned transferable features. The proposed method is verified by two fault diagnosis cases of bearings, in which the health states of locomotive bearings in real cases are identified by using the data respectively collected from motor bearings and gearbox bearings in laboratories. The results show that the proposed method is able to effectively learn transferable features to bridge the discrepancy between the data from BLMs and BRMs. Consequently, it presents higher diagnosis accuracy for BRMs than existing methods.
As one of most critical component of high-speed locomotive, wheel set bearing fault identification has attracted an increasing attention in recent years. However, non-stationary vibration signal with ...modulation phenomenon and heavy background noise make it difficult to excavate the hidden weak fault feature. Variational Mode Decomposition (VMD), which can decompose the non-stationary signal into couple Intrinsic Mode Functions adaptively and non-recursively, brings a feasible tool. However, heavy background noise seriously affects setting of mode number, which may lead to information loss or over decomposition problem. In this paper, an independence-oriented VMD method via correlation analysis is proposed to adaptively extract weak and compound fault feature of wheel set bearing. To overcome the information loss problem, the appropriate mode number is determined by the criterion of approximate complete reconstruction. Then the similar modes are combined according to the similarity of their envelopes to solve the over decomposition problem. Finally, three applications to wheel set bearing fault of high speed locomotive verify the effectiveness of the proposed method compared with original VMD, EMD and EEMD methods.
•Advantage and limitation of variational mode decomposition on signal feature extraction are studied.•Independence-oriented VMD is proposed for solving information loss and over decomposition problem orderly to identify wheel set bearing fault.•Experimental validation and engineering applications are carried out to demonstrate the feasibility of proposed method.
Multi-modal displays that allow the locomotive engineer to delay safety-critical dispatches in high workload scenarios offer the promise of reducing the cognitive distraction that occurs when the ...locomotive engineer must listen to a dispatcher’s communication. In an effort to determine whether locomotive engineers could delay safety-critical information from the dispatcher in high workload scenarios, we developed and evaluated such a multi-modal display system. It was hypothesized that locomotive engineers, when provided with the ability to postpone the delivery of information from the dispatcher, would perform better than locomotive engineers who were not provided that capability. Contrary to the above hypothesis, an analysis of the eye tracking measures indicated that the engineers performed more poorly in the multi-modal display system condition, indicating that the system as designed did not allow the engineer to safely delay dispatch messages. We conclude that aspects of the new system that seemed to increase distraction should be redesigned to modify how and when the engineer uses the system to access information and allow for a safe delay of safety-critical information.
•Without the prior period and the choice of the order of shift, IMCKD can enhance fault feature more effectively than MCKD.•Omitting the resampling process, IMCKD is more appropriate in the diagnosis ...of rolling bearing fault.•IMCKD expands the application to rolling bearing compound-fault diagnosis.•The proposed method is validated by both simulated and experimental data.
The extraction of periodic impulses, which are the important indicators of rolling bearing faults, from vibration signals is considerably significance for fault diagnosis. Maximum correlated kurtosis deconvolution (MCKD) developed from minimum entropy deconvolution (MED) has been proven as an efficient tool for enhancing the periodic impulses in the diagnosis of rolling element bearings and gearboxes. However, challenges still exist when MCKD is applied to the bearings operating under harsh working conditions. The difficulties mainly come from the rigorous requires for the multi-input parameters and the complicated resampling process. To overcome these limitations, an improved MCKD (IMCKD) is presented in this paper. The new method estimates the iterative period by calculating the autocorrelation of the envelope signal rather than relies on the provided prior period. Moreover, the iterative period will gradually approach to the true fault period through updating the iterative period after every iterative step. Since IMCKD is unaffected by the impulse signals with the high kurtosis value, the new method selects the maximum kurtosis filtered signal as the final choice from all candidates in the assigned iterative counts. Compared with MCKD, IMCKD has three advantages. First, without considering prior period and the choice of the order of shift, IMCKD is more efficient and has higher robustness. Second, the resampling process is not necessary for IMCKD, which is greatly convenient for the subsequent frequency spectrum analysis and envelope spectrum analysis without resetting the sampling rate. Third, IMCKD has a significant performance advantage in diagnosing the bearing compound-fault which expands the application range. Finally, the effectiveness and superiority of IMCKD are validated by a number of simulated bearing fault signals and applying to compound faults and single fault diagnosis of a locomotive bearing.
•A highly accurate intelligent scheme is proposed for fault diagnosis.•The combination between an adaptive learning rate method and Nesterov momentum is proposed to improve the standard deep belief ...network.•The proposed method can automatically extract valid fault features from frequency domain signals.•A new adaptive learning rate deep belief network with Nesterov momentum capable of identifying different fault states is constructed.
The effective fault diagnosis of rotating machinery is critical to ensure the continuous operation of equipment and is more economical than scheduled maintenance. Traditional signal processing-based and artificial intelligence-based methods, such as wavelet packet transform and support vector machine, have been proved effective in fault diagnosis of rotating machinery, which prevents unexpected machine breakdowns due to the failure of significant components. However, these methods have several disadvantages that make them unable to automatically and effectively extract valid fault features for the effective fault diagnosis of rotating machinery. A novel adaptive learning rate deep belief network combined with Nesterov momentum is developed in this study for rotating machinery fault diagnosis. Nesterov momentum is adopted to replace traditional momentum to enable declining in advance and to improve training performance. Then, an individual adaptive learning rate method is used to select a suitable step length for accelerating descent. To confirm the utility of the proposed deep learning network architecture, two examinations are implemented on datasets from gearbox and locomotive bearing test rigs. Results indicate that the method achieves impressive performance in fault pattern recognition. Comparisons with existing methods are also conducted to demonstrate that the proposed method is more accurate and robust.
Eukaryotic cells migrate by coupling the intracellular force of the actin cytoskeleton to the environment. While force coupling is usually mediated by transmembrane adhesion receptors, especially ...those of the integrin family, amoeboid cells such as leukocytes can migrate extremely fast despite very low adhesive forces
. Here we show that leukocytes cannot only migrate under low adhesion but can also transmit forces in the complete absence of transmembrane force coupling. When confined within three-dimensional environments, they use the topographical features of the substrate to propel themselves. Here the retrograde flow of the actin cytoskeleton follows the texture of the substrate, creating retrograde shear forces that are sufficient to drive the cell body forwards. Notably, adhesion-dependent and adhesion-independent migration are not mutually exclusive, but rather are variants of the same principle of coupling retrograde actin flow to the environment and thus can potentially operate interchangeably and simultaneously. As adhesion-free migration is independent of the chemical composition of the environment, it renders cells completely autonomous in their locomotive behaviour.
In order to efficiently absorb more regenerative braking energy which sustains much longer compared with the conventional vehicle, and guarantee the safety of the hybrid system under the actual ...driving cycle of locomotive, an energy management control based on dynamic factor strategy is proposed for a scale-down locomotive system which consists of proton exchange membrane fuel cell (PEMFC) and battery pack. The proposed strategy which has self-adaption function for different driving cycles aims to achieve the less consumption of hydrogen and higher efficiency of the hybrid system. The experimental results demonstrate that the proposed strategy is able to maintain the charge state of battery (SOC) better than Equivalent Consumption Minimization Strategy (ECMS), and the proposed strategy could keep the change trend of SOC, which the final SOC is closed to the target value regardless of the initial SOC of battery. Moreover, the hydrogen consumption has been reduced by 0.86g and the efficiency of overall system has been raised of 2% at least than ECMS under the actual driving cycle through the proposed strategy. Therefore, the proposed strategy could improve the efficiency of system by diminishing the conversion process of energy outputted by fuel cell.
•Proposing an energy management strategy based on dynamic power factor for fuel cell/battery hybrid locomotive.•Describing and establishing a scale-down experiment platform of fuel cell hybrid locomotive system.•Analyzing the efficiency and fuel consumption of each parts of system.
Developing biomimetic cartilaginous tissues that support locomotion while maintaining chondrogenic behavior is a major challenge in the tissue engineering field. Specifically, while locomotive forces ...demand tissues with strong mechanical properties, chondrogenesis requires a soft microenvironment. To address this challenge, 3D cartilage‐like tissue is fabricated using two biomaterials with different mechanical properties: a hard biomaterial to reflect the macromechanical properties of native cartilage, and a soft biomaterial to create a chondrogenic microenvironment. To this end, a bath composed of an interpenetrating polymer network (IPN) of polyethylene glycol (PEG) and alginate hydrogel (MPa order compressive modulus) is developed as an extracellular matrix (ECM) with self‐healing properties. Within this bath supplemented with thrombin, human mesenchymal stem cell (hMSC) spheroids embedded in fibrinogen are 3D bioprinted, creating a soft microenvironment composed of fibrin (kPa order compressive modulus) that simulate cartilage's pericellular matrix and allow a fast diffusion of nutrients. The bioprinted hMSC spheroids present high viability and chondrogenic‐like behavior without adversely affecting the macromechanical properties of the tissue. Therefore, the ability to locally bioprint a soft and cell stimulating biomaterial inside of a mechanically robust hydrogel is demonstrated, thereby uncoupling the micro‐ and macromechanical properties of the 3D printed tissues such as cartilage.
In this work, 3D bioprinting technology is used to develop a biomimetic cartilage‐like tissue with near‐paradoxical mechanical properties, being soft at the cellular level, due to the soft bioink composed of human bone marrow mesenchymal stem cells in the form of spheroids embedded in fibrinogen, and the stiff polyethylene glycol and alginate bath, showing great potential for cartilage regeneration studies.
•Envelope harmonic-to-noise ratio is proposed for periodic impulses detection.•EHNR has intrinsic properties suitable for bearing diagnostics.•The EHNR-based method is proposed to locate the periodic ...impulses in frequency domain.•The EHNR-based method has better performance than kurtogram.
Rolling element bearings are one of the fundamental and most important elements in machines and their failures are among the foremost frequent causes of machine breakdown. Vibration and acoustic signals from faulty bearings are typically a mixture of fault-induced periodic impulses and other components. Traditional time-domain features like root-mean-square (RMS) and kurtosis fail to utilize the periodicity property of the impulses, which makes them invalid in some circumstance. Impulses occurring at specific period is the key characteristic of corresponding defect. In the paper, a novel feature named envelope harmonic-to-noise ratio (EHNR) is proposed for periodic impulses detection. The properties of EHNR are illustrated by simulations and bearing full life cycle degradation data. Moreover, an EHNR-based method is proposed to locate periodic impulses in frequency domain. A simulation and a locomotive bearing test rig are used to verify the proposed method. The proposed method has better performances than kurtosis-based method.
PDF
