Author Center
Editor Work
Peer Review
Editorial Entry
Email Alert
RSS2021 Vol. 40, No. 7
Display Method:
2021, 40(7): 985-992.
doi: 10.13433/j.cnki.1003-8728.20200177
PDF 3241KB
Abstract:
To solve the problems of traditional prosthetic knees such as nonadjustable damping, low imitation and high price, a magnetorheological damper-based prosthetic knee (MRPK) using magnetorheological damper (MRD) as control device is designed. Based on the gait data of healthy people walking on the ground, the structural design of the uniaxial MRPK was carried out, and its dynamic model in the swing phase was established. The damping force required by MRD is greater than 208.6 N and the stroke is greater than 33.3 mm. The structure of the MRD was designed and the simulation of electromagnetic field of the designed MRD was performed with the ANSYS. At the same time, the forward and reverse mechanical models of the MRD were established and its performance curves were obtained. The control system of the designed MRPK was established with the co-simulation method, and the computing torque plus PD (CT+PD) trajectory tracking controller was designed and compared with the PD controller. The results show that the maximum error of the CT+PD controller is -4.6°, while that of the PD controller is 12.3°, therefore proving that the CT+PD trajectory tracking control is effective for the swing control of the designed MRPK.
To solve the problems of traditional prosthetic knees such as nonadjustable damping, low imitation and high price, a magnetorheological damper-based prosthetic knee (MRPK) using magnetorheological damper (MRD) as control device is designed. Based on the gait data of healthy people walking on the ground, the structural design of the uniaxial MRPK was carried out, and its dynamic model in the swing phase was established. The damping force required by MRD is greater than 208.6 N and the stroke is greater than 33.3 mm. The structure of the MRD was designed and the simulation of electromagnetic field of the designed MRD was performed with the ANSYS. At the same time, the forward and reverse mechanical models of the MRD were established and its performance curves were obtained. The control system of the designed MRPK was established with the co-simulation method, and the computing torque plus PD (CT+PD) trajectory tracking controller was designed and compared with the PD controller. The results show that the maximum error of the CT+PD controller is -4.6°, while that of the PD controller is 12.3°, therefore proving that the CT+PD trajectory tracking control is effective for the swing control of the designed MRPK.
2021, 40(7): 993-999.
doi: 10.13433/j.cnki.1003-8728.20200168
Abstract:
In order to study the effects of polymer on the coalescence and separation efficiency of oil droplets in hydrocyclone, the paper studies a spiral-conducting inner cone hydrocyclone object and uses the population balance model (PBM) to study the conditions of no polymer or the 0.5‰ polymer concentration. The velocity field, viscosity field, oil droplets coalescence, migration characteristics and separation efficiency of the oil-water phase in the hydrocyclone are numerically analyzed and verified by experiments. The analysis results indicate that the addition of polymer increases viscosity of the water phase, the maximum residence time and axial motion distance of the oil droplets in the hydrocyclone, thus increasing the chance of coalescence of oil droplets but reducing the axial speed of the oil droplets moving upward at the axial center, making oil drops not flow out quickly from the overflow and finally flowing out from the downflow which increases the difficulty of oil-water separation. Compared with the condition of no polymer, the particle size of the oil droplets and the oil volume fraction at the axial center of the hydrocyclone are higher, but the simplified separation efficiency decreases from 99.79% to 94.72% on the condition that the polymer concentration is 0.5‰. The oil droplet sizes at the inlets and outlets of the two cases are measured by the Marvin particle size meter. The oil droplet sizes at the overflow and downflow on the condition that the polymer concentration is 0.5‰ is higher than those without polymer, verifying the accuracy of the simulation results.
In order to study the effects of polymer on the coalescence and separation efficiency of oil droplets in hydrocyclone, the paper studies a spiral-conducting inner cone hydrocyclone object and uses the population balance model (PBM) to study the conditions of no polymer or the 0.5‰ polymer concentration. The velocity field, viscosity field, oil droplets coalescence, migration characteristics and separation efficiency of the oil-water phase in the hydrocyclone are numerically analyzed and verified by experiments. The analysis results indicate that the addition of polymer increases viscosity of the water phase, the maximum residence time and axial motion distance of the oil droplets in the hydrocyclone, thus increasing the chance of coalescence of oil droplets but reducing the axial speed of the oil droplets moving upward at the axial center, making oil drops not flow out quickly from the overflow and finally flowing out from the downflow which increases the difficulty of oil-water separation. Compared with the condition of no polymer, the particle size of the oil droplets and the oil volume fraction at the axial center of the hydrocyclone are higher, but the simplified separation efficiency decreases from 99.79% to 94.72% on the condition that the polymer concentration is 0.5‰. The oil droplet sizes at the inlets and outlets of the two cases are measured by the Marvin particle size meter. The oil droplet sizes at the overflow and downflow on the condition that the polymer concentration is 0.5‰ is higher than those without polymer, verifying the accuracy of the simulation results.
2021, 40(7): 1000-1008.
doi: 10.13433/j.cnki.1003-8728.20200166
Abstract:
Due to the lack of theoretical guidance in differential local mean decomposition (DLMD) method and the inability of traditional performance degradation index to accurately represent the current state of rolling bearing in the whole life stage, a new rolling bearing performance evaluation method based on Hilbert-differential local mean decomposition (HDLMD) and Jensen-Renyi divergence (JRD) was proposed. Firstly, the original vibration signal is decomposed by HDLMD and the product function matrix (PF) is extracted. Then, the PF component containing the most fault information is selected based on Laplacian score. Then calculate the probability distribution of the effective PF component after filtering, and obtain the Renyi entropy value of the effective PF component. Finally, the JRD distance between normal signal and signal of different fault degree is calculated, and the degenerate state of rolling bearing is judged. The rolling bearing experimental data from Case Western Reserve University and the life-cycle data from the National Aeronautics and Space Administration show that the proposed method can accurately and effectively evaluate the state of bearing performance degradation.
Due to the lack of theoretical guidance in differential local mean decomposition (DLMD) method and the inability of traditional performance degradation index to accurately represent the current state of rolling bearing in the whole life stage, a new rolling bearing performance evaluation method based on Hilbert-differential local mean decomposition (HDLMD) and Jensen-Renyi divergence (JRD) was proposed. Firstly, the original vibration signal is decomposed by HDLMD and the product function matrix (PF) is extracted. Then, the PF component containing the most fault information is selected based on Laplacian score. Then calculate the probability distribution of the effective PF component after filtering, and obtain the Renyi entropy value of the effective PF component. Finally, the JRD distance between normal signal and signal of different fault degree is calculated, and the degenerate state of rolling bearing is judged. The rolling bearing experimental data from Case Western Reserve University and the life-cycle data from the National Aeronautics and Space Administration show that the proposed method can accurately and effectively evaluate the state of bearing performance degradation.
2021, 40(7): 1009-1015.
doi: 10.13433/j.cnki.1003-8728.20200457
Abstract:
Aiming at the problem that the traditional null space obstacle avoidance methods cannot take the obstacle avoidance behavior in advance according to the obstacle distance, while ensure the end-effector tracking accuracy, a trajectory tracking algorithm of null space obstacle avoidance for end-effector of manipulators is proposed in this paper. In this algorithm, the pseudo-distance instead of Euclidean distance is first used as the proximity for distance to complete null space obstacle avoidance for manipulators. Then, an adaptive positive definite matrix K and velocity error saturation function sat( ė ) are designed to feedback the results of the real-time trajectory tracking to the inverse kinematics of redundant manipulators; next, according to the results of feedback, the joint angular velocity is adaptively adjusted to minimize the trajectory tracking error of end-effector. The iiwa14 manipulator was employed for the simulation experiment, and the simulation results showed that the proposed algorithm can accomplish the null space obstacle avoidance for redundant manipulator while ensuring the end-effector trajectory tracking error below 1 cm, which verifies the effectiveness and superiority of the proposed algorithm.
Aiming at the problem that the traditional null space obstacle avoidance methods cannot take the obstacle avoidance behavior in advance according to the obstacle distance, while ensure the end-effector tracking accuracy, a trajectory tracking algorithm of null space obstacle avoidance for end-effector of manipulators is proposed in this paper. In this algorithm, the pseudo-distance instead of Euclidean distance is first used as the proximity for distance to complete null space obstacle avoidance for manipulators. Then, an adaptive positive definite matrix K and velocity error saturation function sat( ė ) are designed to feedback the results of the real-time trajectory tracking to the inverse kinematics of redundant manipulators; next, according to the results of feedback, the joint angular velocity is adaptively adjusted to minimize the trajectory tracking error of end-effector. The iiwa14 manipulator was employed for the simulation experiment, and the simulation results showed that the proposed algorithm can accomplish the null space obstacle avoidance for redundant manipulator while ensuring the end-effector trajectory tracking error below 1 cm, which verifies the effectiveness and superiority of the proposed algorithm.
2021, 40(7): 1016-1024.
doi: 10.13433/j.cnki.1003-8728.20200182
Abstract:
Early fault feature of rolling bearings is very weak and affected by environmental noise, which makes its signal-to-noise ratio very low and makes it difficult to extract the weak fault feature. To solve this problem, this paper put forward an integrated diagnosis method based on the adaptive local iterative filter (ALIF) and maximum correlated kurtosis deconvolution (MCKD). Firstly the collected vibration signals were decomposed by ALIF into a number of narrow band intrinsic mode functions (IMFs), and two sensitive IMF components screened out according to the correlation coefficient -kurtosis criterion are reconstructed for noise reduction. Then, the MCKD algorithm is used to enhance the shock component of the fault characteristics. Finally, the envelope spectrum demodulation analysis was carried out for the signal enhanced by the application of ALIF-MCKD and the fault characteristics were extracted to determine the bearing fault location.
Early fault feature of rolling bearings is very weak and affected by environmental noise, which makes its signal-to-noise ratio very low and makes it difficult to extract the weak fault feature. To solve this problem, this paper put forward an integrated diagnosis method based on the adaptive local iterative filter (ALIF) and maximum correlated kurtosis deconvolution (MCKD). Firstly the collected vibration signals were decomposed by ALIF into a number of narrow band intrinsic mode functions (IMFs), and two sensitive IMF components screened out according to the correlation coefficient -kurtosis criterion are reconstructed for noise reduction. Then, the MCKD algorithm is used to enhance the shock component of the fault characteristics. Finally, the envelope spectrum demodulation analysis was carried out for the signal enhanced by the application of ALIF-MCKD and the fault characteristics were extracted to determine the bearing fault location.
2021, 40(7): 1025-1030.
doi: 10.13433/j.cnki.1003-8728.20200190
Abstract:
Aiming at the problem of autonomous selection and optimization of process parameters of robot grinding and polishing systems, an optimization method of the processing parameters in the grinding and polishing based on neural networks and genetic algorithms is proposed. Adopting artificial neural networks based workpiece surface roughness prediction model to solve complex non-linear problems among the processing parameters. Combining the roughness prediction model with the grinding and polishing efficiency formula, by using genetic algorithms to globally optimize each processing parameter to solve the dual-objective optimization problem of processing quality and efficiency and finally obtain the optimal processing parameter combination. On the premise of meeting the requirements of processing quality, the processing efficiency has been improved by nearly one third, which proves that this processing parameter optimization method is feasible and effective.
Aiming at the problem of autonomous selection and optimization of process parameters of robot grinding and polishing systems, an optimization method of the processing parameters in the grinding and polishing based on neural networks and genetic algorithms is proposed. Adopting artificial neural networks based workpiece surface roughness prediction model to solve complex non-linear problems among the processing parameters. Combining the roughness prediction model with the grinding and polishing efficiency formula, by using genetic algorithms to globally optimize each processing parameter to solve the dual-objective optimization problem of processing quality and efficiency and finally obtain the optimal processing parameter combination. On the premise of meeting the requirements of processing quality, the processing efficiency has been improved by nearly one third, which proves that this processing parameter optimization method is feasible and effective.
2021, 40(7): 1031-1036.
doi: 10.13433/j.cnki.1003-8728.20200173
Abstract:
In order to make the globoidal CAM mechanism in the design process has the modification flexibility, the design cycle being short and efficient and so on, this paper uses NX Open C classic application programming interfaces (apis) and Microsoft visual studio (VS) to realize the connection, so that the user can employ the NX Open application development menus, toolbar, dialog boxes, such as tools to implement and interaction design, dialog box by calling the callback function to activate the corresponding controls, application (callback function) through the appropriate programming language and NX code Open API. Based on the secondary development function of NX computer software, a parameterization system suitable for rapid modeling of globoidal indexing CAM mechanism was developed in combination with VS to realize visual interaction design and global optimization. The results show that the system with parameterization design greatly shortens the design time and reduces the tedious repetitive work.
In order to make the globoidal CAM mechanism in the design process has the modification flexibility, the design cycle being short and efficient and so on, this paper uses NX Open C classic application programming interfaces (apis) and Microsoft visual studio (VS) to realize the connection, so that the user can employ the NX Open application development menus, toolbar, dialog boxes, such as tools to implement and interaction design, dialog box by calling the callback function to activate the corresponding controls, application (callback function) through the appropriate programming language and NX code Open API. Based on the secondary development function of NX computer software, a parameterization system suitable for rapid modeling of globoidal indexing CAM mechanism was developed in combination with VS to realize visual interaction design and global optimization. The results show that the system with parameterization design greatly shortens the design time and reduces the tedious repetitive work.
2021, 40(7): 1037-1042.
doi: 10.13433/j.cnki.1003-8728.20200191
Abstract:
In order to study the particle flow field distribution and movement characteristics under different amplitudes, frequencies and barrel widths in one-dimensional vibratory barrel finishing, the numerical simulation and analysis of the barrel finishing process are carried out based on discrete element method. The study finds that the particle group in processing is divided into upper, middle and lower layers according to the position of the particles. The particle velocity of the upper layer fluctuates greatly, and after a period of time, the particle velocity will be in a dynamic equilibrium state; the particle velocity of the middle and bottom layers is always constant fluctuation in the range and periodicity; particles in different layers will climb after they start to move, and they will be in dynamic equilibrium after a period of time. The upper layer particles have the largest climb value and the bottom layer particles have the smallest climb value. As the frequency and amplitude increase, the climbing height of the particles increases; the wider the tube width, the smaller the climbing height.
In order to study the particle flow field distribution and movement characteristics under different amplitudes, frequencies and barrel widths in one-dimensional vibratory barrel finishing, the numerical simulation and analysis of the barrel finishing process are carried out based on discrete element method. The study finds that the particle group in processing is divided into upper, middle and lower layers according to the position of the particles. The particle velocity of the upper layer fluctuates greatly, and after a period of time, the particle velocity will be in a dynamic equilibrium state; the particle velocity of the middle and bottom layers is always constant fluctuation in the range and periodicity; particles in different layers will climb after they start to move, and they will be in dynamic equilibrium after a period of time. The upper layer particles have the largest climb value and the bottom layer particles have the smallest climb value. As the frequency and amplitude increase, the climbing height of the particles increases; the wider the tube width, the smaller the climbing height.
2021, 40(7): 1043-1050.
doi: 10.13433/j.cnki.1003-8728.20200178
Abstract:
Aiming at the current situation that Computer numerical control(CNC) tools in the manufacturing workshop are prone to overuse or replacement in the course of continuous operation, the method for obtaining tool wear perception data and the establishment of wear prediction model framework are studied. In order to avoid the influence of the sensors noise, OPC technology is used to directly cooperate with the machine tool to complete the CNC communication, and a set of dual-lens vertical distribution of the perceptual data acquisition system is designed; in order to enhance the generalization ability of the prediction model, Dropout optimized deep belief network (DBN) is used to establish the prediction model, the optimal weights are reconstructed at the feature extraction stage, and then the feature matching stage is introduced for training the tag amount. The results show that the average accuracy of the improved DBN algorithms prediction is about 96.0%, which is significantly improved comparing with the prediction accuracy and reliability of the traditional models.
Aiming at the current situation that Computer numerical control(CNC) tools in the manufacturing workshop are prone to overuse or replacement in the course of continuous operation, the method for obtaining tool wear perception data and the establishment of wear prediction model framework are studied. In order to avoid the influence of the sensors noise, OPC technology is used to directly cooperate with the machine tool to complete the CNC communication, and a set of dual-lens vertical distribution of the perceptual data acquisition system is designed; in order to enhance the generalization ability of the prediction model, Dropout optimized deep belief network (DBN) is used to establish the prediction model, the optimal weights are reconstructed at the feature extraction stage, and then the feature matching stage is introduced for training the tag amount. The results show that the average accuracy of the improved DBN algorithms prediction is about 96.0%, which is significantly improved comparing with the prediction accuracy and reliability of the traditional models.
2021, 40(7): 1051-1057.
doi: 10.13433/j.cnki.1003-8728.20200172
Abstract:
Aiming at the problems such as component failure and rolling bearing fault in the electric spindle of machining centers, an improved DBN(Deep belief network) electric spindle fault diagnosis method is proposed in this paper. The method extracts features from vibration signals of rolling bearings in the electric spindle under running fault conditions, and then maps the complex relationship between the signal and the fault feature through the DBN to perform fault diagnosis. In order to improve the efficiency of training DBN and solve the problem of gradient disappearance during backpropagation, a new activation function is established. The research results show that the DBN with the new activation function not only reduces the time cost, but also has the higher ability of fault identification.
Aiming at the problems such as component failure and rolling bearing fault in the electric spindle of machining centers, an improved DBN(Deep belief network) electric spindle fault diagnosis method is proposed in this paper. The method extracts features from vibration signals of rolling bearings in the electric spindle under running fault conditions, and then maps the complex relationship between the signal and the fault feature through the DBN to perform fault diagnosis. In order to improve the efficiency of training DBN and solve the problem of gradient disappearance during backpropagation, a new activation function is established. The research results show that the DBN with the new activation function not only reduces the time cost, but also has the higher ability of fault identification.
2021, 40(7): 1058-1064.
Abstract:
As a typical hard-brittle material, ZrO2 ceramics is difficult to obtain good surface quality with ordinary grinding methods. However, ultrasonic vibration grinding can significantly improve its machining effect. In this paper, longitudinal torsion ultrasonic vibration is applied to grinding, and the comparison test of the ordinary grinding (OG) with the longitudinal torsion ultrasonic grinding (LTUG) was designed with single factor method. The surface roughness Ra value and microstructure of processed materials were used as evaluation indexes. The influence law of each technological parameter on the surface quality is analyzed and obtained. The result shows that the surface Ra value via LTUG was always lower than that via OG, and the grinding surface was more uniform and smooth; comparing with OG, Ra value via LTUG decreases firstly and then increases with the increasing of ultrasonic energy, and at the same time, they both decrease firstly and then increase with the increasing of grinding depth, and decrease gradually with the increasing of spindle speed; in addition, the surface Ra value via OG increases firstly and then decreases with the increasing of feed speed, while the LTUG increases continuously and gradually approaches OG.
As a typical hard-brittle material, ZrO2 ceramics is difficult to obtain good surface quality with ordinary grinding methods. However, ultrasonic vibration grinding can significantly improve its machining effect. In this paper, longitudinal torsion ultrasonic vibration is applied to grinding, and the comparison test of the ordinary grinding (OG) with the longitudinal torsion ultrasonic grinding (LTUG) was designed with single factor method. The surface roughness Ra value and microstructure of processed materials were used as evaluation indexes. The influence law of each technological parameter on the surface quality is analyzed and obtained. The result shows that the surface Ra value via LTUG was always lower than that via OG, and the grinding surface was more uniform and smooth; comparing with OG, Ra value via LTUG decreases firstly and then increases with the increasing of ultrasonic energy, and at the same time, they both decrease firstly and then increase with the increasing of grinding depth, and decrease gradually with the increasing of spindle speed; in addition, the surface Ra value via OG increases firstly and then decreases with the increasing of feed speed, while the LTUG increases continuously and gradually approaches OG.
2021, 40(7): 1065-1071.
doi: 10.13433/j.cnki.1003-8728.20200176
Abstract:
The non-circular gear sector of the automobile steering gear rocker shaft is a complex structure gear whose displacement coefficient of the upper cross section of the tapered surface changes with different end cross section positions. There is currently no dedicated measurement software. This paper presents a method for measuring the tooth profile error of the non-circular gear sector of the rocker arm of the steering gear of an automobile. The theoretical tooth profile coordinates calculated by the tooth profile normal method are imported into the Sprocket profile module in Kligenberg P26 to measure non-circular gear sector with known theoretical tooth profile coordinates. According to the meshing characteristics of the non-circular gear sector rack pair, the instantaneous center of meshing of the non-circular gear sector rack pair is calculated and drawn; the tooth profile of the non-circular gear sector according to the Willis theorem is calculated, and the tooth profile by using the simulation software to is drawn; the error of the tooth surface of the non-circular tooth is intuitively displayed, the tooth surface mesh is divided, and several sections of the same tooth groove is integrated into a complete tooth surface error model.
The non-circular gear sector of the automobile steering gear rocker shaft is a complex structure gear whose displacement coefficient of the upper cross section of the tapered surface changes with different end cross section positions. There is currently no dedicated measurement software. This paper presents a method for measuring the tooth profile error of the non-circular gear sector of the rocker arm of the steering gear of an automobile. The theoretical tooth profile coordinates calculated by the tooth profile normal method are imported into the Sprocket profile module in Kligenberg P26 to measure non-circular gear sector with known theoretical tooth profile coordinates. According to the meshing characteristics of the non-circular gear sector rack pair, the instantaneous center of meshing of the non-circular gear sector rack pair is calculated and drawn; the tooth profile of the non-circular gear sector according to the Willis theorem is calculated, and the tooth profile by using the simulation software to is drawn; the error of the tooth surface of the non-circular tooth is intuitively displayed, the tooth surface mesh is divided, and several sections of the same tooth groove is integrated into a complete tooth surface error model.
2021, 40(7): 1072-1078.
doi: 10.13433/j.cnki.1003-8728.20200169
Abstract:
In the method of variable prediction pattern recognition (VPMCD), which is based on the internal relationship between the eigenvalues to establish the prediction model, the traditional recognition method is greatly affected by the outliers predicted by individual features in the eigenvector, which is easy to cause classification errors. In this paper, a discriminant function (RWESOS) based on the least square sum of error weighted by ratio is proposed, which can greatly reduce the feature weight of abnormal prediction, increase the weight of correct prediction features, and improve the classification accuracy. Experiments show that the recognition rate of proposed RWESOS-VPMCD method using RWESOS discriminant function is 4% and 11% higher than that of BP neural network and common discriminant function respectively.
In the method of variable prediction pattern recognition (VPMCD), which is based on the internal relationship between the eigenvalues to establish the prediction model, the traditional recognition method is greatly affected by the outliers predicted by individual features in the eigenvector, which is easy to cause classification errors. In this paper, a discriminant function (RWESOS) based on the least square sum of error weighted by ratio is proposed, which can greatly reduce the feature weight of abnormal prediction, increase the weight of correct prediction features, and improve the classification accuracy. Experiments show that the recognition rate of proposed RWESOS-VPMCD method using RWESOS discriminant function is 4% and 11% higher than that of BP neural network and common discriminant function respectively.
2021, 40(7): 1079-1084.
doi: 10.13433/j.cnki.1003-8728.20200181
Abstract:
To overcome the disadvantages of limited motion range and easy damage to objects, this paper proposes a novel micro-gripper integrated with bridge type amplification mechanism and lever mechanism. The gripper can adapt to different size objects avoiding damage or detachment of tiny objects during micro-clamping operations. Following issues were addressed in this paper, design of the micro-gripper described in detail, established the mathematical model according to the working principle of the micro-gripper arm, and obtained the magnification ratio of the micro-gripper arm. In addition, static and dynamic simulations were performed using the finite element analysis software ANSYS Workbench, including the verified motion range. The results showed that the theoretical calculation value of the magnification ratio is consistent with the simulation analysis value.
To overcome the disadvantages of limited motion range and easy damage to objects, this paper proposes a novel micro-gripper integrated with bridge type amplification mechanism and lever mechanism. The gripper can adapt to different size objects avoiding damage or detachment of tiny objects during micro-clamping operations. Following issues were addressed in this paper, design of the micro-gripper described in detail, established the mathematical model according to the working principle of the micro-gripper arm, and obtained the magnification ratio of the micro-gripper arm. In addition, static and dynamic simulations were performed using the finite element analysis software ANSYS Workbench, including the verified motion range. The results showed that the theoretical calculation value of the magnification ratio is consistent with the simulation analysis value.
2021, 40(7): 1085-1090.
doi: 10.13433/j.cnki.1003-8728.20200184
Abstract:
Bearings are an important component of wind turbines, and their health directly affects the stability of wind turbine operation and the reliability of the work site. Due to the special working environment of wind turbines, the collected vibration signals contain a large amount of noise interference, it is difficult to accurately extract the information contained in the bearing vibration signals, and also hard to assess the health status of the main bearing. Therefore, in this paper, the Fourier decomposition method (FDM) and Stochastic resonance (SR) are combined to extract the weak bearing vibration information in the signals. Firstly, FDM is used to adaptively decompose the original signals into a series of Fourier intrinsic band functions (FIBFS), which containing bearing vibration characteristics, then find the FIBFS with large correlation for reconstruction, and finally use SR to analyze the reconstructed signals to obtain the characteristic frequency. The results show that the combination of the two methods can effectively improve the output signal to noise ratio, improve the accuracy of the characteristic frequency detection, and provide help for the early diagnosis of weak faults in wind turbine bearings.
Bearings are an important component of wind turbines, and their health directly affects the stability of wind turbine operation and the reliability of the work site. Due to the special working environment of wind turbines, the collected vibration signals contain a large amount of noise interference, it is difficult to accurately extract the information contained in the bearing vibration signals, and also hard to assess the health status of the main bearing. Therefore, in this paper, the Fourier decomposition method (FDM) and Stochastic resonance (SR) are combined to extract the weak bearing vibration information in the signals. Firstly, FDM is used to adaptively decompose the original signals into a series of Fourier intrinsic band functions (FIBFS), which containing bearing vibration characteristics, then find the FIBFS with large correlation for reconstruction, and finally use SR to analyze the reconstructed signals to obtain the characteristic frequency. The results show that the combination of the two methods can effectively improve the output signal to noise ratio, improve the accuracy of the characteristic frequency detection, and provide help for the early diagnosis of weak faults in wind turbine bearings.
2021, 40(7): 1091-1099.
doi: 10.13433/j.cnki.1003-8728.20200170
Abstract:
There exists some problems in traditional data-driven fault diagnosis methods, for example, it is difficult to adaptively extract effective features from bearing signals, cannot make full use of the timing characteristics of fault data, and lacks of the ability to adaptively process dynamic information. An intelligent bearing fault diagnosis method combined deep convolutional neural network and long-term and short-term memory network is proposed. This method constructed a kind of deep networks and can adaptively extract the robust features from the original bearing signals, and then utilized the long short-term memory network to learn the time-dependent relationship in these features, and achieved high-accuracy bearing fault diagnosis. The proposed method overcame the problems existed in the traditional feature extraction methods, such as heavy dependence on expert experience and incomplete utilization for time series information, and realized intelligent and accurate diagnosis of faults. The experimental results show that the proposed method can extract more accurate features and make the fault diagnosis more intelligent and reliable by utilizing the timing information in the process of fault degradation
There exists some problems in traditional data-driven fault diagnosis methods, for example, it is difficult to adaptively extract effective features from bearing signals, cannot make full use of the timing characteristics of fault data, and lacks of the ability to adaptively process dynamic information. An intelligent bearing fault diagnosis method combined deep convolutional neural network and long-term and short-term memory network is proposed. This method constructed a kind of deep networks and can adaptively extract the robust features from the original bearing signals, and then utilized the long short-term memory network to learn the time-dependent relationship in these features, and achieved high-accuracy bearing fault diagnosis. The proposed method overcame the problems existed in the traditional feature extraction methods, such as heavy dependence on expert experience and incomplete utilization for time series information, and realized intelligent and accurate diagnosis of faults. The experimental results show that the proposed method can extract more accurate features and make the fault diagnosis more intelligent and reliable by utilizing the timing information in the process of fault degradation
2021, 40(7): 1100-1105.
doi: 10.13433/j.cnki.1003-8728.20200185
Abstract:
In order to improve the bearing capacity of high-speed electric multiple units (EMU) motor hangers, considering the randomness of geometric parameters, the response surface method is combined with multi-objective genetic algorithm to optimize its reliability. Firstly, the static strength and displacement sensitivity analysis of the motor hanger are finished to find the design variables; secondly, the central combination test design is carried out for the design variables, and the polynomial response surface model of static strength and displacement is fitted according to the test design points; then, the randomness of geometric parameters is transformed into reliability, and the least square method is used to fit the reliability function equation as the constraint condition; finally, the multi-objective genetic algorithm is used to optimize the reliability of the motor hanger. The research results show that the combination of response surface method and structural reliability optimization method can simplify the solution process and improve the reliability of optimization results.
In order to improve the bearing capacity of high-speed electric multiple units (EMU) motor hangers, considering the randomness of geometric parameters, the response surface method is combined with multi-objective genetic algorithm to optimize its reliability. Firstly, the static strength and displacement sensitivity analysis of the motor hanger are finished to find the design variables; secondly, the central combination test design is carried out for the design variables, and the polynomial response surface model of static strength and displacement is fitted according to the test design points; then, the randomness of geometric parameters is transformed into reliability, and the least square method is used to fit the reliability function equation as the constraint condition; finally, the multi-objective genetic algorithm is used to optimize the reliability of the motor hanger. The research results show that the combination of response surface method and structural reliability optimization method can simplify the solution process and improve the reliability of optimization results.
2021, 40(7): 1106-1113.
doi: 10.13433/j.cnki.1003-8728.20200171
Abstract:
The damping characteristic test of the magnetorheological damper used in the suspension system is performed. The Levenberg-Marquardt optimization algorithm is used to identify the parameters of the adjustable Sigmoid model of the magnetorheological damper. The identified parameters are fitted using the least square method. A sliding mode controller for a quarter-vehicle suspension system is designed; the switching surface parameters are determined using the pole placement method. To alleviate the chattering problem of the sliding mode control system, the saturation function is used instead of the symbolic function. Fuzzy control and RBF neural network are used to optimize sliding mode controller of semi-active suspension. With the input of a random road excitation, the simulation analysis of the semi-active suspension with the optimized sliding mode controller is performed. Simulation results show that the adjustable Sigmoid model can achieve a more precise control of the damping force, and this model identified by the optimization algorithm can control the quarter-vehicle suspension system well. Moreover, RBF optimized sliding mode controller has better performance than fuzzy sliding mode controller.
The damping characteristic test of the magnetorheological damper used in the suspension system is performed. The Levenberg-Marquardt optimization algorithm is used to identify the parameters of the adjustable Sigmoid model of the magnetorheological damper. The identified parameters are fitted using the least square method. A sliding mode controller for a quarter-vehicle suspension system is designed; the switching surface parameters are determined using the pole placement method. To alleviate the chattering problem of the sliding mode control system, the saturation function is used instead of the symbolic function. Fuzzy control and RBF neural network are used to optimize sliding mode controller of semi-active suspension. With the input of a random road excitation, the simulation analysis of the semi-active suspension with the optimized sliding mode controller is performed. Simulation results show that the adjustable Sigmoid model can achieve a more precise control of the damping force, and this model identified by the optimization algorithm can control the quarter-vehicle suspension system well. Moreover, RBF optimized sliding mode controller has better performance than fuzzy sliding mode controller.
2021, 40(7): 1114-1119.
doi: 10.13433/j.cnki.1003-8728.20200188
Abstract:
The hybrid vehicle transmission system is subjected to multiple sources of excitation such as the engine, motor, transmission components, and road surface excitation, and a large dynamic load generates when the mode is switched, which induces complex torsional vibration problems and directly affects the smoothness, reliability and safety of the transmission system. In order to suppress the torsional vibration of the transmission system, an active control strategy based on the hybrid adaptive algorithms for the pure electric mode of the transmission system is proposed. The state-space method was used to establish the dual-mass kinematic balance equation of the transmission system, and the active vibration damping control model for the transmission electric mode was constructed by combining the feedforward controller and feedback controller. By analyzing the effect of the ideal model damping ratio on the control effect, the optimal damping ratio of the transmission system was obtained. Through simulation analysis, the control effects with and without interference are compared. The results show that the hybrid adaptive control algorithm proposed in this paper can reduce the torsional vibration of the drive train by 80%~90% and advance the system stability time by 83%~87%. It effectively suppresses the torsional vibration of the HEV when it starts in pure electric mode, and can eliminate external interference at a specific frequency.
The hybrid vehicle transmission system is subjected to multiple sources of excitation such as the engine, motor, transmission components, and road surface excitation, and a large dynamic load generates when the mode is switched, which induces complex torsional vibration problems and directly affects the smoothness, reliability and safety of the transmission system. In order to suppress the torsional vibration of the transmission system, an active control strategy based on the hybrid adaptive algorithms for the pure electric mode of the transmission system is proposed. The state-space method was used to establish the dual-mass kinematic balance equation of the transmission system, and the active vibration damping control model for the transmission electric mode was constructed by combining the feedforward controller and feedback controller. By analyzing the effect of the ideal model damping ratio on the control effect, the optimal damping ratio of the transmission system was obtained. Through simulation analysis, the control effects with and without interference are compared. The results show that the hybrid adaptive control algorithm proposed in this paper can reduce the torsional vibration of the drive train by 80%~90% and advance the system stability time by 83%~87%. It effectively suppresses the torsional vibration of the HEV when it starts in pure electric mode, and can eliminate external interference at a specific frequency.
2021, 40(7): 1120-1127.
doi: 10.13433/j.cnki.1003-8728.20200167
Abstract:
Inserting adhesive in the steel-aluminum clinched joint can improve the joint performance. The mechanical behavior and failure mechanism of the joint are very complicated. In this paper, the steel-aluminum clinch-bonded hybrid joint of the dual-phase high-strength steel DP590 and aluminum alloy 6061-T6 was manufactured. And the mechanical behavior and failure mechanism were studied with the simulation and experimental methods. Based on ABAQUS finite element analysis software, the mixed failure model for GTN (Gurson tvergaard needleman) + cohesive zone (CZ) was used to simulate the forming and failure process of the clinch-bonded hybrid joint. Through the shear test of the clinched, bonded and clinch-bonded joints, the failure mode and mechanical behavior of the three joints were analyzed. Combining with the adhesive distribution and the failure process of the adhesive layer, the failure mechanism of the hybrid joint was revealed.
Inserting adhesive in the steel-aluminum clinched joint can improve the joint performance. The mechanical behavior and failure mechanism of the joint are very complicated. In this paper, the steel-aluminum clinch-bonded hybrid joint of the dual-phase high-strength steel DP590 and aluminum alloy 6061-T6 was manufactured. And the mechanical behavior and failure mechanism were studied with the simulation and experimental methods. Based on ABAQUS finite element analysis software, the mixed failure model for GTN (Gurson tvergaard needleman) + cohesive zone (CZ) was used to simulate the forming and failure process of the clinch-bonded hybrid joint. Through the shear test of the clinched, bonded and clinch-bonded joints, the failure mode and mechanical behavior of the three joints were analyzed. Combining with the adhesive distribution and the failure process of the adhesive layer, the failure mechanism of the hybrid joint was revealed.
2021, 40(7): 1128-1136.
doi: 10.13433/j.cnki.1003-8728.20200174
Abstract:
Crystallizer is the core component of continuous casting, which is called the "heart" of continuous casting. The length of the paste zone and the location of the initial freezing point are the key factors to evaluate whether the crystallizer can successfully pull out the ingot. In this paper, MILE algorithm, Mix of Eulerian and Lagrangian, in the finite element analysis software ProCAST is used to simulate the temperature field of crystallizers with the heights of 20 mm, 30 mm and 40 mm under the same technological conditions. Through numerical simulation analysis, it is found that the crystallizer with a height of 30 mm is better than that with the heights of 20 mm and 40 mm, with a larger range of adjustable process parameters, and more safe and stable production of ingots. Then, the crystallizer with a height of 30 mm was tested and verified, and the results were in a good agreement with the simulation. And the rationality and superiority of crystallizer design by simulation analysis are proved.
Crystallizer is the core component of continuous casting, which is called the "heart" of continuous casting. The length of the paste zone and the location of the initial freezing point are the key factors to evaluate whether the crystallizer can successfully pull out the ingot. In this paper, MILE algorithm, Mix of Eulerian and Lagrangian, in the finite element analysis software ProCAST is used to simulate the temperature field of crystallizers with the heights of 20 mm, 30 mm and 40 mm under the same technological conditions. Through numerical simulation analysis, it is found that the crystallizer with a height of 30 mm is better than that with the heights of 20 mm and 40 mm, with a larger range of adjustable process parameters, and more safe and stable production of ingots. Then, the crystallizer with a height of 30 mm was tested and verified, and the results were in a good agreement with the simulation. And the rationality and superiority of crystallizer design by simulation analysis are proved.
2021, 40(7): 1137-1142.
doi: 10.13433/j.cnki.1003-8728.20200461
Abstract:
In order to study the effect of the impact velocity and temperature on the compression after impact of laminates, a progressive damage model considering temperature effect is established to predict the damage mode of laminates, and the damage mechanism of laminates under low velocity impact is analyzed. Firstly, the displacement-stress curve predicted by the model is compared with the experimental. The influence of the different impact velocity on the indentation depth, delamination area and residual strength is further analyzed. Finally, the influence of the different temperature on the residual strength and delamination area at an impact velocity of 2.3 m/s is analyzed.
In order to study the effect of the impact velocity and temperature on the compression after impact of laminates, a progressive damage model considering temperature effect is established to predict the damage mode of laminates, and the damage mechanism of laminates under low velocity impact is analyzed. Firstly, the displacement-stress curve predicted by the model is compared with the experimental. The influence of the different impact velocity on the indentation depth, delamination area and residual strength is further analyzed. Finally, the influence of the different temperature on the residual strength and delamination area at an impact velocity of 2.3 m/s is analyzed.
2021, 40(7): 1143-1148.
doi: 10.13433/j.cnki.1003-8728.20200442
Abstract:
The influence of the electric heating heat load on helicopter windshield has been researched under extreme flight conditions in this paper. A method of applying non-uniform load aerodynamic loads on the surface of windshield model has been established by parametric processing of discrete aerodynamic load. Forced states of the helicopter windshield in each flight attitude were analyzed in the finite element software ABAQUS and the dangerous flight conditions were found. Electro-thermal model of the windshield was established in ANSYS, the temperature distribution in the thickness direction of the windshield during the electric heating of the windshield in a low temperature environment has been calculated. The thermal-mechanical coupling model of the electrically heated windshield was established to analyze the stress and deformation of windshield in simulated flight conditions. The results showed that electric heating has a positive effect on reducing the tensile stress on the edge of the windshield in a low temperature environment, but has no obvious effect on the deformation of the windshield.
The influence of the electric heating heat load on helicopter windshield has been researched under extreme flight conditions in this paper. A method of applying non-uniform load aerodynamic loads on the surface of windshield model has been established by parametric processing of discrete aerodynamic load. Forced states of the helicopter windshield in each flight attitude were analyzed in the finite element software ABAQUS and the dangerous flight conditions were found. Electro-thermal model of the windshield was established in ANSYS, the temperature distribution in the thickness direction of the windshield during the electric heating of the windshield in a low temperature environment has been calculated. The thermal-mechanical coupling model of the electrically heated windshield was established to analyze the stress and deformation of windshield in simulated flight conditions. The results showed that electric heating has a positive effect on reducing the tensile stress on the edge of the windshield in a low temperature environment, but has no obvious effect on the deformation of the windshield.
