Article

Remaining useful life prediction of circuit breaker operating mechanisms based on wavelet-enhanced dual-tree residual networks

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

The remaining useful life prediction of circuit breaker operating mechanisms is crucial for the condition-based maintenance of national power grids. To realize accurate remaining useful life prediction, a novel wavelet-enhanced dual-tree residual network is proposed in this paper. Through this wavelet transform, the time series is decomposed into two components (high frequency and low frequency). Then the two decomposed components are fed into two lightweight residual neural network structures. By concatenating the dual-tree features, the remaining useful life of a circuit breaker operating mechanism can be predicted. The proposed network is validated using a full-life cycle experiment of the circuit breaker operating mechanism. Results show that the proposed method has good capability when it comes to predicting the remaining useful life of the circuit breaker operating mechanism. Along with application in the construction of smart grids and green energy, it is expected that the proposed method has potential in running state prognostics of circuit breakers.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... It detected, dynamics of various faults through feature extraction, feature selection and signal denoising and predicted them by using the adaptive Bayesian algorithm, and the results showed that the method had a high prediction accuracy (Rezamand et al., 2020). In order to achieve accurate prediction of remaining lifespan, Wu et al. (2024) designed a residual lifespan prediction model based on wavelet enhanced dual tree residual network. The model decomposed time series through wavelet transform and predicted remaining lifespan by concatenating dual tree features. ...
... The model decomposed time series through wavelet transform and predicted remaining lifespan by concatenating dual tree features. The results showed that the prediction effect of that method was good (Wu et al., 2024). Li et al. (2023) researchers designed a deep adversarial network-based residual service life prediction method for partial sensor failure to achieve a good electromechanical health assessment, which extracted generalized sensor invariant features through adversarial learning to make a full use of the information from different sensors, and the findings indicated that the method had a high robustness. ...
... Some researchers have drawn stress relaxation curves of compression springs through stress relaxation tests to indirectly reflect spring performance characteristics and identify the operating status of circuit breakers, but this method requires disassembling the circuit breaker springs [9,10]. Traditional HVCBs fault diagnosis usually involves monitoring signals, such as coil current, vibration signals, travel curves, and acoustic signals, combined with algorithms to detect circuit breaker performance [11][12][13][14][15]. The coil current of HVCBs contains rich information, but current signal analysis is mostly used to find defects in secondary circuits and can only reflect about 25% of mechanical faults in HVCBs [16,17]. ...
Article
Full-text available
Diagnosing the operational status of High‐voltage circuit breakers (HVCBs) is crucial for ensuring the safe and stable operation of the grid. Mechanical characteristic parameters are effective indicators for evaluating the performance of HVCBs. Recent studies have shown that the actions of the springs and cams in HVCBs can be used to detect the operational status of the mechanical mechanisms, which occur extremely quickly, usually in the speed of m/ms. In this paper, dynamic vision sensing technology was employed to rapidly and dynamically capture the movements of the springs and cam of the HPL245B1 HVCB. The data volume of a single experiment is less than 100 MB, whereas the data collected by a high‐speed camera at the same frame rate exceeds 1 GB. Action data streams of the springs and cam were obtained and images were reconstructed from the event streams. The Lucas–Kanade optical flow algorithm and the normalised cross‐correlation algorithm are applied to calculate the parameters of spring deformation characteristics and cam rotation characteristics for mechanical feature detection of HVCBs. This is the first attempt to utilize brain‐inspired hardware technology for the status monitoring of electrical equipment. The advantages of dynamic vision sensing technology, such as high dynamic range, low data transmission, and low energy consumption, also offer significant benefits for air discharge monitoring and status monitoring of electrical equipment.
... With the rapid development of machine learning technology, artificial intelligence (AI) based fault diagnosis and prediction have increasingly become an important strategy for equipment safety and service monitoring 12 . Via related intelligent algorithms, the data-driven diagnostic method can adaptively identify equipment operation status information from existing data without the need of prior knowledge for professional technicians 13 . With an edge-labeling graph neural network method, Zhi et al. 14 propose a tool for wear condition monitoring using wear images which suitable for small sample conditions. ...
Article
Full-text available
Rapid tool wear conditions during the manufacturing process are crucial for the enhancement of product quality. As an extension of our recent works, in this research, a generic in-situ tool wear condition monitoring during the end milling process based on dynamic mode and abnormal evaluation is proposed. With the engagement of dynamic mode decomposition, the real-time response of the sensing physical quantity during the end milling process can be predicted. Besides, by constructing the graph structure of the time series and calculating the difference between the predicted signal and the real-time signal, the anomaly can be acquired. Meanwhile, the tool wear state during the end milling process can be successfully evaluated. The proposed method is validated in milling tool wear experiments and received positive results (the mean relative error is recorded as 0.0507). The research, therefore, paves a new way to realize the in-situ tool wear condition monitoring.
Article
Full-text available
Finite control set model predictive control (FCS-MPC) stands out for fast dynamics and easy inclusion of multiple nonlinear control objectives. However, for long horizontal prediction or complex topologies with multiple levels and phases, the required computation burden surges exponentially as the increases of candidate switch states during one control period. This phenomenon leads to longer sample period to guarantee enough time for traverse progress of cost function minimization. In other words, the allowed highest switching frequency is bounded considerably far from the physical limits, especially for wide-band semiconductor applications. To overcome this issue, the parallel computing characteristic of artificial neural network (ANN) motivates the idea of an ANN-based FCS-MPC imitator (ANN-MPC). In this article, ANN-MPC is implemented on a neutral point clamped (NPC) converter using a shallow neural network. The expert (FCS-MPC) is initially designed, and the basic structure, including activation function selection, training data generation, and offline training progress, and online operation of the imitator (ANN-MPC) are then discussed. After the design of the expert and imitator, a comparative analysis is conducted by field programmable gate array (FPGA) in-the-loop implementation in MATLAB/Simulink environment. The verification results of ANN-MPC show highly similarly qualified control performance and considerably reduced computation resource requirement.
Article
Full-text available
Uncertainty dimensions of geometrical features in film cooling holes will inevitably affect the aerothermal behavior and mechanical characteristics of the engine. To realize the measurement of key parameters of film cooling holes, this article introduces a vision-based method for dimensional in situ measurement of the holes in aero-engines during the laser beam drilling process. In the measurement process, images of the holes can be acquired through the equipped camera in the laser drilling machine. Specifically, dual-tree complex wavelet transform is applied to eliminate the overwhelming interfering noise and preserves the necessary edge information; a local Gini index-based method is proposed to extract the edge information from the complex texture contained in the workpiece surface. Furthermore, the least square approach is employed for dimension measurement. Besides, a nickel-based wafer drilling experiment is presented on a femtosecond five-axis laser drilling machine. Compared with the off-line vision method, the experiment results indicate the calculated mean absolute errors of the diameter and the roundness of the proposed method are evaluated to be 0.00005 mm and 0.01113 mm respectively. This study, therefore, paves the way for in situ measurement of film cooling holes in aero-engines without any additional measuring instrument during the laser drilling process.
Article
Full-text available
Remaining useful life (RUL) prediction can effectively avoid unexpected mechanical breakdowns, thus improving operational reliability. However, the distribution discrepancy caused by different working conditions may lead to deterioration in the prognostic task of machinery. Inspired by the idea of transfer learning, a novel intelligent approach based on dynamic domain adaptation (DDA) is proposed for the machinery RUL prediction of multiple working conditions in this paper. At first, reverse validation technology is utilized to select appropriate source samples to construct the training dataset. Then two dynamic domain adaptation networks are trained to extract domain invariant degradation feature and predict RUL, namely dynamic distribution adaptation network and dynamic adversarial adaptation network. In the dynamic domain adaptation network, the fuzzy set theory is employed to calculate conditional distribution discrepancy loss, and the dynamic adaptive factor is introduced to dynamically adjust the distribution weights. Finally, the proposed method is proved to be effective through two run-to-failure bearing datasets. Related experimental results indicate that, compared with other related RUL prediction methods, the DDA-based prognostic method not only achieves better prediction performance, but also avoids the influence of negative transfer and distribution weight fluctuation.
Article
Full-text available
Robust bearing fault detection is significant to reduce the machinery down-time, and to prevent catastrophic failure. Many algorithms are proposed for the faults feature extraction, but it remains challenging to monitors the condition of the mechanical systems from the overwhelming interference noise contained signal in a short response time. To address this problem, as an extension of our recent work, this paper introduces an enhanced framework using acquired time-series signals. Specifically, an improved Hankel matrix-based method is proposed for the identification of the state from the sampled vibration signal for each spindle turn, where matrix similarity is employed for the mechanical operation state monitoring. The experimental results indicate that the proposed method performs considerably well in fault identification (100% identification accuracy in three tests) even with few data samples and phase shift. This work therefore would have more hopeful prospects in a variety of engineering fault detection applications.
Conference Paper
Full-text available
Data-driven techniques, especially on artificial intelligence (AI) such as deep learning (DL) techniques, have attracted more and more attention in the manufacturing sector because of the growth of industrial Internet of Things (IoT) and Big Data. Tremendous researches of DL techniques have been applied in machine health monitoring, but still very limited works focus on the application of DL on the Remaining Useful Life (RUL) prediction. Precise RUL prediction can significantly improve the reliability and operational safety of the industrial components or systems, avoid fatal breakdown and reduce the maintenance costs. This paper gives a brief introduction of RUL prediction and reviews the start-of-the-art DL approaches in terms of four main representative deep architectures, including Auto-encoder, Deep Belief Network (DBN), Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN). It has been observed that DL techniques attract growing interests on RUL prediction that suggests a promising future of their applications in manufacturing.
Article
Full-text available
Machined surfaces are rough from a microscopic perspective no matter how finely they are finished. Surface roughness is an important factor to consider during production quality control. Using modern techniques, surface roughness measurements are beneficial for improving machining quality. With optical imaging of machined surfaces as input, a convolutional neural network (CNN) can be utilized as an effective way to characterize hierarchical features without prior knowledge. In this paper, a novel method based on CNN is proposed for making intelligent surface roughness identifications. The technical scheme incorporates there elements: texture skew correction, image filtering, and intelligent neural network learning. Firstly, a texture skew correction algorithm, based on an improved Sobel operator and Hough transform, is applied such that surface texture directions can be adjusted. Secondly, two-dimensional (2D) dual tree complex wavelet transform (DTCWT) is employed to retrieve surface topology information, which is more effective for feature classifications. In addition, residual network (ResNet) is utilized to ensure automatic recognition of the filtered texture features. The proposed method has verified its feasibility as well as its effectiveness in actual surface roughness estimation experiments using the material of spheroidal graphite cast iron 500-7 in an agricultural machinery manufacturing company. Testing results demonstrate the proposed method has achieved high-precision surface roughness estimation.
Article
Robust spring energy state identification of the operating mechanism is of great significance for monitoring the overall performance of the circuit breakers. However, rapid monitoring of the spring energy storage state based on the acquired current signal during the service period has not yet been realized. To address this problem, this research put forward a hybrid method for spring energy storage state identification and successfully applied it to the operating mechanism of circuit breakers. In this method, the Gramian angular field (GAF) is employed to represent the dynamic characteristics evolution process. Furthermore, combined with a convolutional block attention module (CBAM) and residual network (ResNet), a hybrid method is proposed for identifying the spring energy storage state and finally verified in the circuit breaker experiment. Experimental results proved the extraordinary efficiency of the proposed method (the average F1 -score is reported as 0.994). The research suggests that the use of GAF might provide a viable source for state identification of operating mechanisms in circuit breakers.
Article
Flexible direct current (DC) grids face a serious challenge in terms of rapidly isolating DC faults. A DC circuit breaker is an effective solution for DC fault isolation. To improve the fault-isolation and reclosing capability of flexible DC systems, a new high voltage direct current (HVDC) circuit breaker topology with adaptive reclosing capability is proposed in this paper. The topology of the circuit breaker is a T-shaped structure, which has the ability to break the current in both directions and effectively reduce the cost of components. Meanwhile, after the fault is cleared, the circuit breaker is controlled to inject a voltage signal into the line. Based on this, to prevent the circuit breaker reclosing in the event of a permanent fault from having an impact on the system, a method for fault identification based on Euclidean distance that uses the voltage signal to identify the fault properties is proposed. Finally, the performance of the circuit breaker is simulated and verified by PSCAD/EMTDC simulation software, and compared with the typical existing circuit breakers to verify the effectiveness of the circuit breaker and reclosing scheme.
Article
The remaining useful life (RUL) prognostic of lithium-ion batteries (LIBs) is important in the reliability of electric vehicles. The degradation state of LIBs is related to the current moment and historical data, which is a non-Markovian process with long-term dependence. This manuscript proposes a RUL prognostic approach based on a non-Markovian process, which uses a fractional Brownian motion (FBM) model. Firstly, a nonlinear FBM model is established to describe the battery non-Markovian capacity fading process. The drift parameter of the FBM model and degradation states are updated by an online Kalman filter when a new measurement value arrives. Then the maximum likelihood estimation approach is introduced to obtain the other undecided fixed parameters. This approach is based on off-line battery degradation historical data. According to the first hitting time, the probability distribution function is derived to quantify the uncertainty of the RUL prognostic results. Finally, two datasets are used to verify the effectiveness of the proposed method. For the NASA dataset battery #5, the relative errors of the RUL prediction results of the proposed method are 2.941 and 2.083 when the starting points of the predictions are 60 cycles and 80 cycles, respectively. Thus, the proposed method is superior to other methods.
Article
Bearing is one of the most important component of rotary machine, and its health state is directly related to the safety of industrial production. In this paper, health state assessment of bearing is investigated with feature enhancement and prediction error compensation. Specially, health state assessment consists of time-to-start prediction point detection and remaining useful life (RUL) prediction. In the first stage, variance feature based on Kalman filter is introduced to detect the time-to-start prediction point. Subsequently, complete ensemble empirical mode decomposition with adaptive noise is employed to reconstruct the degradation trend, and cumulative function is adopted to realize the feature enhancement, then efficient health indicator can be constructed. In the RUL prediction stage, degradation model and adaptive extended Kalman filter are fused to achieve the prediction, and bidirectional gated recurrent unit neural network is chosen to compensate the prediction error. Finally, experimental studies based on PRONOSTIA and XJTU-SY datasets are conducted to validate the effectiveness of the proposed method and its superiority over the traditional methods.
Article
Wavelet-based techniques are strongly recommended as a good alternative for the fast detection and characterization of voltage sags. However, the accuracy and effectiveness of these techniques greatly depend on selecting an appropriate mother wavelet. Therefore, in this work, a wavelet correlation-based technique has been developed to select the most appropriate mother wavelet for the characterization and detection of voltage sag. The efficacy and accuracy of the proposed method are tested with twenty different mother wavelets on various voltage sag signals, namely, recorded industrial, multi-stage, and synthetic signals under different conditions of unbalanced. It is shown that the mother wavelet having the highest similarity with a voltage sag provides the best results for its characterization and detection. Further, the various performance parameters of voltage sags, namely magnitude, duration, sag initiation, recovery, are evaluated with the proposed method and results are compared with Independent Component Analysis (ICA), hybrid wavelet, dq-transformation, Enhanced Phase Locked Loop (EPLL), Fast Fourier Transform (FFT) methods showing that the performance of the proposed method is better than other existing methods for sag detection. In addition, the proposed method can also be used for estimating the magnitude of voltage sags.
Article
Deep learning methods have achieved noteworthy seeing results in the mechanical fault diagnosis of high-voltage circuit breakers with the recent advancements in artificial intelligence. However, the premise of the above method for obtaining excellent performance is to have sufficient samples, which is impractical due to the characteristics of the high-voltage circuit breakers. This study proposes a novel U-Net with CapsNet for high-voltage circuit breakers fault diagnosis to resolve these issues, achieving a high-precision and robust diagnosis of few-shot high-voltage circuit breakers. In a few-shot diagnosis, the U-Net with CapsNet takes advantage of the high accuracy of the U-Net. The capsule network is used in the contraction and expansion paths of the original U-Net to reduce the loss of features in the pooling process. The forward transfer from the bottom to the high-level capsule is completed by the dynamic routing algorithm. The feature information in the high-level capsule is unified with the bottom layer. The experimental results show that using the U-Net with CapsNet proposal, we can quickly and accurately realize the fault diagnosis of few-shot high-voltage circuit breakers, with an accuracy of 93.25%. The model has faster convergence speed and better stability, which provides a reliable solution for efficient and accurate fault diagnosis of high-voltage circuit breakers compared with traditional methods.
Article
Multi-energy complementary integrated energy system (MCIES) is considered as a promising solution to mitigate carbon emissions and promote carbon peaking and carbon neutrality. Currently, the capacities of a MCIES are sized according to the deterministic load and parameters of the system model. However, uncertainty may lead to the failure to achieve the desired performance and affect the sizing of the MCIES. This study explored an optimization model for the proper sizing of the MCIES considering uncertainties to achieve the best economic, environmental and thermal comfort benefits. The non-dominated sorting genetic algorithm-II (NSGA-II) combined with technique for order preference by similarity to an ideal solution (TOPSIS) and Shannon entropy method were adopted to solve the optimization. Case studies, an actual swimming pool building with MCIES, as the prototype, were used to illustrate the procedure. Moreover, the effects of uncertainty degree and scenario setting were investigated. The results show the benefits of the proposed approach against the traditional deterministic optimization method for comprehensive consideration of economy, environment and thermal comfort. It also suggests that uncertainty and scenario setting should be careful and proper consideration during the design stage, as they have a significant impact on the results of sizing.
Article
Accurate prediction of remaining useful life (RUL) is necessary to ensure stable and safe operations for rocket engines. The paper proposed a multi-head attention network coupled with adaptive meta-transfer learning for RUL prediction. By combining the convolution-based branch with an attention-based branch, the multi-head attention network is proposed for accurate RUL prediction of cryogenic bearings in rocket engines under the steady stage. In addition, an adaptive model-agnostic meta-transfer learning strategy is developed to further improve the performance under small sample circumstances with adaptive hyper-parameters. To demonstrate the superiority, the proposed method is compared with typical benchmark algorithms using real monitoring data from a high-precision cryogenic rocket engine experiment platform. Results indicate that the proposed method achieves better performance compared with existing models under multiple evaluation indexes.
Article
The interturn faults of voltage transformers will cause inaccurate data collection, wrong decision and wrong action of relay protection in power system. It is necessary to put forward an effective on method about the voltage transformer assessment. Two novel sensitive methods based on the excitation current and the phase difference between primary voltage and the excitation current (phase difference) for identifying interturn faults during energization of voltage transformer are presented. The kirchhoff’s voltage equation (KVE) are provided by calculating the excitation current and phase difference between primary voltage and excitation current (phase difference). The proposed method was analyzed during the simulations and experimental tests were also carried out in laboratory to validate the results. The results indicate that the severity of interturn fault is positively correlated with the excitation current, and negatively correlated with the phase difference. Furthermore, the phase difference can effectively detect early interturn fault of voltage transformer.
Article
Traditional deep learning methods do not effectively extract degradation features from vibration signals widely used for remaining useful life (RUL) prediction while avoiding the gradient problem. This paper proposes a residual neural network framework with multiscale attention mapping (ResNet-MA) based on the vibration signal's characteristics to solve the above problems. In the framework, we first use the decomposed vibration signal processed by the ensemble empirical mode decomposition method (EEMD) as the model's input. To improve the neural network's ability to extract degradation signals, channel attention mapping, time attention mapping, and multiscale pooling methods are used in ResNet-MA. Experimental verification and analysis are carried out with the available data sets, which show that the proposed method's prediction accuracy is 14% higher than the residual neural network of deep learning before the improvement, and 3-6% higher than the state-of-the-art related algorithms.
Article
Multiple modes of vibration are usually incorporated in a single record of vibration measurement in condition monitoring of rotating machinery. Wavelet transform is an effective tool to detect and isolate transient fault features from other interfering modes. The conventional dyadic wavelet transform decomposes the signal into wavelet subspaces with distinct central frequencies and specific frequency bandwidths. In this paper, we propose a novel theory of centralized multiresolution analysis (CMR) and reveal the implicit fractal geometry properties in CMR. A concept of nested centralized wavelet packet space (NCWPS) is introduced to describe the self-similarity phenomenon in CMR. Within the theoretical framework, the classical dyadic wavelet packet is assimilated as a subordinated proper-subset of the augmented NCWPS. Moreover, the generalized CMR characterized by tunable and flexible frequency-scale topology configuration is established using harmonic wavelet transform. The CMR can be regarded as an improved transient signature dictionary. Therefore, the CMR is combined with an improved stationary signature dictionary to ensure enhanced performance in fault feature extraction in multiple modes coupled vibration measurements. The effectiveness of the proposed method is validated using numerical simulations, a rub-impact experiment, and a case study of vibration signal analysis in steel making industry.
Article
This paper presents an integrated approach for the detection and classification of the faults of rolling bearing in rotary machines. Permutation entropy (PE) is integrated with a flexible analytical wavelet transform (FAWT). The signals from healthy and faulty bearing systems with different operating conditions are decomposed by FAWT. PE values of each sub-bands are calculated at different levels. To compare effectiveness of the proposed methodology, dyadic discrete wavelet transform (DWT) is used in integration with PE. Signals are decomposed into a set of approximate and detailed coefficients by DWT. PE values of decomposed signals are calculated and are then fed as feature vectors to support vector machine (SVM) classifier for the classification of different types and fault sizes. The classification results of both approaches are compared. The results demonstrate the effectiveness and robustness of FAWT-integrated-PE over the DWT integrated with PE, for detection of bearing faults and their classification.
Article
The estimation of the remaining useful life (RUL) of rolling element bearings has been an area of excessive research over the past few decades. Time-series forecasting approaches are the most popular methods for calculating the RUL of bearings. However, there exist two key challenges in predicting the RUL using time-series forecasting. The first is the development of an accurate health indicator (HI) that can indicate bearing degradation i.e. the developed HI must be capable of tracking the early and critical degradation stages in bearings. The second is the determination of a suitable failure threshold for the HI time-series. The HIs for different bearings fluctuate at different levels at the time of failure and consequently, it becomes difficult to set a definite failure threshold for them. To overcome these problems, a novel HI is proposed in this paper. First, the vibration signals acquired from the bearings are subjected to the feature extraction process. For this purpose, chaotic features determined using the Lyapunov exponent are utilized. Then, feature samples extracted under healthy bearing conditions are used to train probabilistic self-organizing map (p-SOM) algorithm. Finally, the trained p-SOM model is tested against the monitored feature samples to construct the HI. This HI has a major advantage in that it lies in a specific range from [0-1] and therefore allows for the selection of precise limits for defining bearing failure. Once the HI has been determined, a state-space model is employed to forecast the HI up to pre-set failure thresholds and subsequently compute the RUL of the bearings. The proposed technique is validated on publicly-available benchmark datasets. The experimental results confirm that the suggested HI is effective in predicting damage growth as well as forecasting the RUL of bearings. Further, it outperforms the traditional health indicator i.e. SOM-based-minimum quantization error (MQE).
Article
In the bearings fault detection applications, as the marginal distribution of wavelet transform output is a heavy-tailed bell-shaped function characterized by a larger portion of small wavelet coefficients or even zeros, the traditional feature extraction approaches such as wavelet-energy spectrum and energy spectrum entropy fail to accurately express the statistical feature of wavelet sub bands. In this paper, we propose a novel wavelet-based bearings fault detection approach using wavelet transform and Generalized Gaussian Density (GGD) modeling. A GGD-based feature descriptor is generated from concatenating the statistical parameters of each wavelet sub band estimated by the maximum likelihood method. According to the descriptor information, a class label for the bearing fault detection is assigned by the subsequent classifier. The extensive experimental results show that the proposed approach can more accurately and flexibly capture wavelet sub band information of bearings vibration signals than energy, energy entropy, Gaussian and Laplace based feature description methods. Moreover, the experimental results with different wavelet filters, decomposition levels, and classifiers demonstrate that the new method significantly improves the fault detection accuracy compared with traditional approaches with better robustness.
Article
Rolling bearings are key components of rotating machinery. Thus the prediction of remaining useful life (RUL) is vital in condition based maintenance. This paper proposes a new method for RUL prediction of bearings based on time-varying Kalman filter, which can automatically match different degradation stages of bearings and effectively realize the prediction of RUL. The evolution of monitoring data in normal and slow degradation stages is linear trend, and the evolution in accelerated degradation stage is non-linear. So Kalman filter models based on linear and quadratic functions are established respectively. Meanwhile, a sliding window relative error is constructed to adaptively judge the bearing degradation stages. It can automatically switch filter models to process monitoring data at different stages. Then the RUL can be predicted effectively. Two groups of bearing run-to-failure datasets are utilized to demonstrate the feasibility and validity of the proposed method.
Article
The metal surface topology contains abundant information related to the health states of the cutting tool as well as the cutting operation. In this paper, we attempt to adopt 2D digital images of the machined metal surface, acquired via non-contact photo-imaging techniques, as the monitoring media. A Wallis filter based dodging algorithm is applied to cure the uneven contrast phenomenon caused by imperfect lighting illumination. 3D digital models were derived and retrieved from the digital image using a wavelet enhanced Shape from shading (SFS) transform. The minimization based SFS is presented to retrieve the 3D digital surface from the milled workpiece. The dual tree complex wavelet transform is adopted to enhance SFS such that the interfering noise can be suppressed. In the end, quantitative surface roughness indicators are utilized to estimate the surface roughness numerically. A milling cutting experiment of aero-material of aluminum alloy 7075 was carried out to verify the effectiveness of the proposed approach. The comparison results demonstrate that the proposed approach was capable of retrieving 3D surfaces of high precision. With the approach, the digital image emerges as a promising vehicle for machining condition monitoring of CNC machines.
Article
Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers---8x deeper than VGG nets but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.
Article
This paper investigates the dynamic response of a spring-type operating mechanism for a 69 kV SF6 gas insulated circuit breaker. This paper uses the equation of motion to analyze the dynamic response of the spring type operating mechanism. The basic theory behind the equation of motion is briefly introduced and the kinematic coefficients of the links and the centers of gravity of the spring-type operating mechanism are then derived. The non-linear differential equation of motion is then solved by using the fourth-order Runge–Kutta method. From these results, the duration of each operation is then calculated. It is shown that the closing time is 0.114 s and the opening time is 0.078 s which are comparable with those obtained experimentally. In addition, the dynamic response of the output moving contact has also been calculated and comments are made on its kinematic characteristics.
Article
The paper discusses the theory behind the dual-tree transform, shows how complex wavelets with good properties can be designed, and illustrates a range of applications in signal and image processing. The authors use the complex number symbol C in CWT to avoid confusion with the often-used acronym CWT for the (different) continuous wavelet transform. The four fundamentals, intertwined shortcomings of wavelet transform and some solutions are also discussed. Several methods for filter design are described for dual-tree CWT that demonstrates with relatively short filters, an effective invertible approximately analytic wavelet transform can indeed be implemented using the dual-tree approach.
Uniform deceleration design for stepped shock absorber in circuit breaker spring operating mechanism
  • K Wan
  • Y Xi
  • X Wang