A multi-scale convolutional auto-encoder and its application in fault diagnosis of rolling bearings

Aiming at the difficulty of fault identification caused by manual extraction of fault features of rotating machinery,a one-dimensional multi-scale convolutional auto-encoder fault diagnosis model is proposed,based on the standard convolutional auto-encoder.In this model,the parallel convolutional and deconvolutional kernels of different scales are used to extract the features from the input signal and reconstruct the input signal;then the feature map extracted by multi-scale convolutional kernels is used as the input of the classifier;and finally the parameters of the whole model are fine-tuned using labeled data.Experiments on one set of simulation fault data and two sets of rolling bearing fault data are conducted to validate the proposed method.The results show that the model can achieve 99.75%,99.3%and 100%diagnostic accuracy,respectively.In addition,the diagnostic accuracy and reconstruction error of the one-dimensional multi-scale convolutional auto-encoder are compared with traditional machine learning,convolutional neural networks and a traditional convolutional auto-encoder.The final results show that the proposed model has a better recognition effect for rolling bearing fault data.

GA-1DLCNN method and its application in bearing fault diagnosis

Due to the fact that the vibration signal of the rotating machine is one-dimensional and the large-scale convolution kernel can obtain a better perception field, on the basis of the classical convolution neural network model(LetNet-5), one-dimensional large-kernel convolution neural network(1 DLCNN) is designed. Since the hyper-parameters of 1 DLCNN have a greater impact on network performance, the genetic algorithm(GA) is used to optimize the hyper-parameters, and the method of optimizing the parameters of 1 DLCNN by the genetic algorithm is named GA-1 DLCNN. The experimental results show that the optimal network model based on the GA-1 DLCNN method can achieve 99.9% fault diagnosis accuracy, which is much higher than those of other traditional fault diagnosis methods. In addition, the 1 DLCNN is compared with one-dimencional small-kernel convolution neural network(1 DSCNN) and the classical two-dimensional convolution neural network model. The input sample lengths are set to be 128, 256, 512, 1 024, and 2 048, respectively, and the final diagnostic accuracy results and the visual scatter plot show that the effect of 1 DLCNN is optimal.

A method for workpiece surface small-defect detection based on CutMix and YOLOv3

Surface small defects are often missed and incorrectly detected due to their small quantity and unapparent visual features.A method named CSYOLOv3,which is based on CutMix and YOLOv3,is proposed to solve such a problem.First,a four-image CutMix method is used to increase the small-defect quantity,and the process is dynamically adjusted based on the beta distribution.Then,the classic YOLOv3 is improved to detect small defects accurately.The shallow and large feature maps are split,and several of them are merged with the feature maps of the predicted branch to preserve the shallow features.The loss function of YOLOv3 is optimized and weighted to improve the attention to small defects.Finally,this method is used to detect 512×512 pixel images under RTX 2060Ti GPU,which can reach the speed of 14.09 frame/s,and the mAP is 71.80%,which is 5%-10%higher than that of other methods.For small defects below 64×64 pixels,the mAP of the method reaches 64.15%,which is 14%higher than that of YOLOv3-GIoU.The surface defects of the workpiece can be effectively detected by the proposed method,and the performance in detecting small defects is significantly improved.

Rolling bearing performance degradation evaluation by VMD and embedding selection-based NPE

In order to improve the incipient fault sensitivity and stability of degradation index in the rolling bearing performance degradation evaluation process,an embedding selection-based neighborhood preserving embedding(ESNPE)method is proposed.Firstly,the acquired vibration signals are decomposed by variational mode decomposition(VMD),and the singular value and relative energy of each intrinsic mode function(IMF)are extracted to form a high-dimensional feature set.Then,the NPE manifold learning method is used to extract the embedded features in the feature space.Considering the problem that useful embedding information is easily suppressed in NPE,an embedding selection strategy is built based on the Spearman correlation coefficient.The effectiveness of embeddings is measured by the coefficient absolute value,and useful embeddings are preserved in the early stage of bearing degradation by using the first-order difference method.Finally,the degradation index is established using the support vector data description(SVDD)model and bearing performance degradation evaluation is achieved.The proposed method was tested with the whole life experiment data of a rolling bearing,and the result was compared with the feature extraction methods of traditional principal component analysis(PCA)and NPE.The results show that the proposed method is superior in improving the incipient fault sensitivity and stability of the degradation index.

A modified time domain interpolation method for LS channel estimation in OFDM systems

A modified time domain interpolation method is proposed for orthogonal frequency division multiplexing(OFDM)systems to address the problem that time domain interpolation in the least square(LS)channel estimation method based on comb-type pilots cannot choose the pilot spacing flexibly.Firstly,the estimated channel frequency response(CFR)at pilot positions in the frequency domain is obtained by LS channel estimation based on comb-type pilots,and the estimated channel impulse response(CIR)in the time domain is obtained by linear interpolation and inverse fast Fourier transform(IFFT).Secondly,the error of the estimated CIR obtained by linear interpolation is analyzed by theoretical deduction,and a method for correcting it is proposed.Finally,an estimated CFR at all subcarrier positions in the frequency domain is obtained by performing zero padding in the time domain and fast Fourier transform(FFT)on the modified CIR.The simulation results suggest that the proposed method gives similar performance to time domain interpolation,yet it does not need to meet the condition of time domain interpolation that the number of subcarriers must be an integral multiple of pilot spacing to use it.The proposed method allows for flexible pilot spacing,reducing the number of pilots and the consumption of subcarriers used for channel estimation.

Online SOC estimation based on modified covariance extended Kalman filter for lithium batteries of electric vehicles

To offset the defect of the traditional state of charge(SOC)estimation algorithm of lithium battery for electric vehicle and considering the complex working conditions of lithium batteries,an online SOC estimation algorithm is proposed by combining the online parameter identification method and the modified covariance extended Kalman filter(MVEKF)algorithm.Based on the parameters identified on line with the multiple forgetting factors recursive least squares methods,the newly-established algorithm recalculates the covariance in the iterative process with the modified estimation and updates the process gain which is used for the next state estimation to decrease errors of the filter.Experiments including constant pulse discharging and the dynamic stress test(DST)demonstrate that compared with the EKF algorithm,the MVEKF algorithm produces fewer estimation errors and can reduce the errors to 5%at most under the complex charging and discharging conditions of batteries.In the charging process under the DST condition,the EKF produces a larger deviation and lacks stability,while the MVEKF algorithm can estimate SOC stably and has a strong robustness.Therefore,the established MVEKF algorithm is suitable for complex and changeable working conditions of batteries for electric vehicles.

Human-machine compatibility and dynamic analysis of a novel unpowered and self-adaptive shoulder rehabilitation exoskeleton

To reduce the complexity of the configuration and control strategy for shoulder rehabilitation exoskeleton,a 2R1R1P2R serial of shoulder exoskeleton based on gravity balance is proposed.Based on three basic rotatory shoulder joints,an exact kinematic constraint system can be formed between the exoskeleton and the upper arm by introducing a passive sliding pair and a center of glenohumeral(CGH)unpowered compensation mechanism,which realizes the human-machine kinematic compatibility.Gravity balance is used in the CGH compensation mechanism to provide shoulder joint support.Meanwhile,the motion of the compensation mechanism is pulled by doing reverse leading through the arm to realize the kinematic self-adaptive,which decreases control complexity.Besides,a simple and intuitive spring adjustment strategy is proposed to ensure the gravity balance of any prescribed quality.Furthermore,according to the influencing factors analysis of the scapulohumeral rhythm,the kinematic analysis of CGH mechanism is performed,which shows that the mechanism can fit the trajectory of CGH under various conditions.Finally,the dynamic simulation of the mechanism is carried out.Results indicate that the compensation torques are reduced to below 0.22 N·m,and the feasibility of the mechanism is also verified.

Comparative study on tribological properties of isostatic graphite and carbon graphite under dry sliding and water-lubricated conditions

The tribological properties of isostatic graphite and carbon graphite under dry sliding and water lubricated conditions were studied.The friction test was conducted by using a pin-on-disc tribometer.The friction coefficient and the wear rate were employed to evaluate the tribological performances of the two materials,and wear morphology was used to analyze the wear mechanism.The results show that the friction coefficient of the isostatic graphite is larger than that of the carbon graphite under the dry sliding condition,and the wear rate is lower than that of the carbon graphite.Under the water lubricated condition,the friction coefficients and the wear rates of the isostatic graphite decrease obviously.The main wear form of the isostatic graphite is abrasive wear,while the main wear form of the carbon graphite is spalling wear.Finally,the tribological mechanism of the isostatic graphite under dry sliding and water lubricated conditions were systematically analyzed.

Evaluation of remaining useful life for corroded pipeline with finite element simulation and reliability theory

An integrated approach was proposed to evaluate the remaining useful life(RUL)of corroded petroleum pipelines.Two types of failure modes(i.e.,leakage and burst failure)were considered,and the corresponding limit state functions(LSFs)were established with the structural reliability theory.A power-law function was applied to model the growth of corrosion defects,and the effect of external environmental factors on the growth of the pipeline s defect was considered.Moreover,the result was compared with the commonly used linear growth model.After that,a finite element simulation model was established to calculate the burst pressure of the pipeline with corrosion defects,and its accuracy was verified through hydraulic burst test and by comparison with international criteria.On that basis,the probability that the pipeline may fail was calculated with Monte Carlo simulation(MCS)and by considering the LSFs,and the pipeline s RUL was obtained accordingly.Furthermore,sensitivity analysis was conducted to determine the sensitivity parameters for the corrosion and RUL of the pipeline.The results indicate that the radial corrosion rate,wall thickness and working pressure have a great influence on the failure probability of the pipeline.Thus,corresponding measures should be adopted during the operation process of the pipeline to reduce the corrosion rate and increase the wall thickness,so as to prolong the pipeline s RUL.

Nonlinear identification and characterization of structural joints based on vibration transmissibility

In order to investigate the nonlinear characteristics of structural joint,the experimental setup with a jointed mass system is established for dynamic characterization analysis and vibration prediction,and a corresponding nonlinearity identification method is studied.First,the sine-sweep vibration test with different baseexcitation levels areapplied to the structural joint system to study the dominant modal of mass rigid motion.Then,based on t e harmonic balance method principle,t e measured vibration transmissibilities a e utilized for nonlinearity identification using different excitation levels.Experimental results show that nonlinear spring and damping force can be represented by a polynomial order approximation.The identified nonlinear stiffness and damping force can predict the system’s response,and they can reveal t e shifts of resonant frequency or damping due to discontinuity of contact mechanisms within a certain range.

Influence of image data set noise on classification with a convolutional network

To evaluate the influence of data set noise, the network in network(NIN) model is introduced and the negative effects of different types and proportions of noise on deep convolutional models are studied. Different types and proportions of data noise are added to two reference data sets, Cifar-10 and Cifar-100. Then, this data containing noise is used to train deep convolutional models and classify the validation data set. The experimental results show that the noise in the data set has obvious adverse effects on deep convolutional network classification models. The adverse effects of random noise are small, but the cross-category noise among categories can significantly reduce the recognition ability of the model. Therefore, a solution is proposed to improve the quality of the data sets that are mixed into a single noise category. The model trained with a data set containing noise is used to evaluate the current training data and reclassify the categories of the anomalies to form a new data set. Repeating the above steps can greatly reduce the noise ratio, so the influence of cross-category noise can be effectively avoided.

Ride quality evaluation of the soil compactor cab supplemented by auxiliary hydraulic mounts via simulation and experiment

In order to evaluate the ride quality of the soil compactor cab supplemented by the auxiliary hydraulic mounts (AHM), a nonlinear dynamic model of the soil compactor interacting with the off-road deformable terrain is established based on Matlab/Simulink sofware. The power spectral density (PSD) and the weighted root mean square (RMS) of acceleration responses of the vertical driver s seat, the cab s pitch and roll angle are chosen as objective functions in low-frequency range. Experimental investigation is also used to verify the accuracy of the model. The influence of the damping coefficients of the AHM on the cab s ride quality is analyzed, and damping coefficients are then optimized via a genetic algorithm program. The research results show that the cab s rubber mounts added by the AHM clearly improve the ride quality under various operating conditions. Particularly, with the optimal damping coefficients of the front-end mounts c a 1,2 = 1 500 N · s/m and of the rear-end mounts c a 3,4 =2 335 N · s/m, the weighted RMS values of the driver s seat, the cab s pitch and roll angle are reduced by 22.2%, 18.8%, 58.7%, respectively. Under the condition of the vehicle travelling, with the optimal damping coefficients of c a 1,2 = 1 500 N · s/m and c a 3,4 =1 882 N · s/m, the maximum PSD values of the driver s seat, the cab s pitch and roll angle are clearly decreased by 36.7%, 54.7% and 50.6% under the condition of the vehicle working.

Effects of crankpin bearing speed and dimension on engine power

A new method combining the slider-crank mechanism dynamic(SCM)and crankpin bearing(CB)lubrication models is proposed to analyze the effects of CB dimensions and engine speed on the lubrication efficiency and friction power loss(LE-FPL)of an engine.The dynamic and lubrication equations are then solved on the basis of the combined model via an algorithm developed in MATLAB.To enhance the reliability of the research results,the experimental data of combustion gas pressure is applied for simulation.The load bearing capacity(or oil film pressure),friction force,friction coefficient,and eccentricity ratio of the CB are selected as objective functions to evaluate the LE-FPL.The effects of engine speed,bearing width,and bearing radius on the LE-FPL are then evaluated.Results show that reductions in engine speed,bearing width,or bearing radius can decrease the FPL but reduce the LE of the engine and vice versa.In particular,the LE-FPL can effectively be improved by slightly reducing the bearing width and bearing radius or maintaining engine speed at 2000 r/min.

A method for constraining the end effect of EMD based on sequential similarity detection and adaptive filter

Aimed at the problem of the end effect when using empirical mode decomposition(EMD),a method for constraining the end effect of EMD is proposed based on sequential similarity detection and adaptive filter.The method divides the signal into many wavelets,and it changes the initial wavelet length to select the best initial wavelet that has the minimum error and maximum number of matching seed wavelets,and the wavelet slopes are used for pre-matching and secondary matching to speed up the matching speed.Then,folded self-adaptive threshold is used to select multiple seed wavelets,and finally the end waveform is predicted and expanded according to the adaptive filter method.The proposed method is used to analyze the non-stationary nonlinear simulation signal and experimental signal,and it is compared with the mirror extension and RBF extension methods.The orthogonality index and similarity index of the EMD results of the extended signal after the proposed method are better than those of the other methods.The results show that the proposed method can better constrain the end effect,and has certain validity,accuracy and stability in solving the end effect problem.

Molecular dynamics simulations of strain-dependent thermal conductivity of single-layer black phosphorus

Classical molecular dynamics(MD)simulations ae performed to investigate the effects of mechanical strain on the thermal conductivity of single-layer black phosphorus(SLBP)nanoribbons along different directions at room temperature.The results show that the tensile strain afects the thermal conductivity of nanoribbons by changing thephonon density of state(DOS)and mean free path(M FP).The thermal conductivity shows a sharp enhancement with the tensile strain applied along the armchai diection,while it increases slowly with the strain applied along the zigzag diection.This phenomenon cm be mainly explained by effects of the phonon DOS and MFP.The increasing strain along the armchai direction weakens DOS and strengthens MFP clearly.However,when it comes to the increasing strain along the zigzag deection'DOS enliances significantly while MFP decreases slightly.The findings explore the relationship between the tensile strain and the thermal conductivity reasonably and can provide a reliable method to estimate the MFP of black phosphorus.

Calculation method for trajectory following control for autonomous vehicles

The current literature lacks uniform calculation methods for following trajectory control for autonomous vehicles,including the calculation of errors,determination of tracking points,and design of feedforward controllers.Hence,a complete calculation method is proposed to address this gap.First,a control equation in the form of an error is obtained according to the dynamic equation of the vehicle coordinate system and the trajectory following model.Secondly,the deviation of the vehicle state is obtained according to the current vehicle s state and the following control model.Finally,a linear quadratic regulator(LQR)controller with feedforward control is designed according to the characteristics of the dynamic equation.With the proposed LQR,the simulation of computational time,anti-interference,and reliability analysis of the trajectory following control is performed by programming using MATLAB.The simulation outcomes are then compared with the experimental results from the literature.The comparison indicates that the proposed complete calculation method is effective,reliable,and capable of achieving real-time and anti-interference following control performance.The simulation results with or without feedforward control show that the steady-state error is eliminated and that good control performance is obtained by introducing feedforward control.

Welding thermal characteristics analysis with numerical simulation for thin-wall parts assembly under different conditions

In order to analyze the welding thermal characteristics problem,the multiscale finite element(FE)model of T-shape thin-wall assembly structure for different thicknesses and the heat source model are established to emphatically study their welding temperature distributions under different conditions.Simultaneously,different welding technology parameters and welding directions are taken into account,and the fillet weld for different welding parameters is employed on the thin-wall parts.Through comparison analysis,the results show that different welding directions,welding thicknesses and welding heat source parameters have a certain impact on the temperature distribution.Meanwhile,for the thin-wall assembly structure of the same thickness,when the heat source is moving,the greater the moving speed,the smaller the heating area,and the highest temperature will decrease.Therefore,the welding temperature field distribution can be altered by adjusting welding parameters,heat source parameters,welding thickness and welding direction,which is conducive to reducing welding deformation and choosing an appropriate and optimal welding thickness of thin-wall parts and relative welding process parameters,thus improving thin-wall welding structure assembly precision in the actual large-size welding structure assembly process in future.

Availability-based maintenance optimization under unequal inspection period and imperfect repair

Taking the multi-component system as research object, a maintenance optimization model based on the unequal inspection period and imperfect repair is established by considering the requirement of expected availability for improving the system＇s availability. An age reduction factor is used to describe the effect of imperfect repair, and the modelling approach for the unequal inspection period is proposed. Unavailable situations are classified into three kinds of independent cases, and the availability is calculated accordingly. Based on the analysis of the relationship between the unavailable cases and the unequal inspection period, an optimization model under imperfect repair is established to optimize the system＇s expected availability. A case study of a wind turbine is provided, and three key components, i.e. gearbox, generator and spindle, are considered. The optimization results of the unequal inspection period model and the equal inspection period model are compared. The results show that the unequal inspection period model based on availability can update the maintenance plan so as to optimize maintenance activities and improve the system＇s availability.

Compact passing algorithm for signalized intersection management based on vehicular network

To improve the traffic efficiency at signalized intersections, a compact passing algorithm is proposed based on a vehicular network. Its basic principle is that several waiting vehicles after the stop line of the considered intersection should simultaneously start in green periods. Thus, more vehicles can pass the intersection in a green period. Then, the having passed vehicles should follow the planned trajectories to enlarge their longitudinal clearances. Phase timing is not considered in the compact passing algorithm, and therefore, the proposed compact passing algorithm can be combined with other algorithms on phase timing to further improve their performances. Several simulations were designed and performed to verify the performance of the proposed algorithm. The simulation results show that the proposed algorithm can increase the number of completed vehicles and decrease the travel time in the signalized intersections managed by fixed-time and vehicle actuated algorithms, which indicates that the proposed algorithm is effective for improving the traffic efficiency at common signalized intersections.

Analysis of radial stiffness of rubber bush used in dynamic vibration absorber

In order to study the influence of the structural parameters of the rubber bush on its radial stiffness, the constitutive relation of rubber materiel is used to obtain the calculation formula of the dimensionless radial stiffness coefficient. The obtained theoretical result is consistent with previous research results in both long rubber bushes and short rubber bushes. The simulation case was conducted by the finite element method to verify the correctness of the theory. The axial compression experiment was conducted to obtain the parameters needed in the simulation. The result shows that the percentage difference between the theoretical result and the simulation one is only 2.75%. A series of simulations were conducted to compare with previous work, and the largest magnitude of the percentage difference is only about 5%. Finally, the radial stiffness experiment was conducted by using a dynamic vibration absorber, and the influence of the structural parameters of the rubber bush on its radial stiffness is obtained. The result shows that the radial stiffness of the rubber bush increases with the increase in the length and the inner radius, but decreases with the increase in the outer radius.