Articles in press have been peer-reviewed and accepted, which are not yet assigned to volumes /issues, but are citable by Digital Object Identifier (DOI).
Display Method:
2025, 39(1): 1-6.
doi: 10.12299/jsues.24-0031
Abstract:
The kinematics and dynamics modeling method for a six-degree-of-freedom stacking robot using Matlab was proposed, as well as stacking trajectory planning and rhythm analysis methods. The improved D-H method was employed to establish coordinate systems and transformation matrices for robot links, followed by the contruction of a simulation model through robot toolbox in Matlab. Trajectory planning and operational rhythm analysis were conducted according to different pallet patterns and stacking postures, which enabled a comprehensive evaluation of trajectory parameters, rhythm time, efficiency and other parameters. Based on the parameters of the mass, moment of inertia, and center of mass of each link of the designed robot, a dynamics model of the robot was established in Matlab for numerical analysis to obtain the torque curves of each joint. Simulation results demonstrate that the motion characteristics of the robot during palletizing operations can provide critical references for mechanical structure design and key component selection.
The kinematics and dynamics modeling method for a six-degree-of-freedom stacking robot using Matlab was proposed, as well as stacking trajectory planning and rhythm analysis methods. The improved D-H method was employed to establish coordinate systems and transformation matrices for robot links, followed by the contruction of a simulation model through robot toolbox in Matlab. Trajectory planning and operational rhythm analysis were conducted according to different pallet patterns and stacking postures, which enabled a comprehensive evaluation of trajectory parameters, rhythm time, efficiency and other parameters. Based on the parameters of the mass, moment of inertia, and center of mass of each link of the designed robot, a dynamics model of the robot was established in Matlab for numerical analysis to obtain the torque curves of each joint. Simulation results demonstrate that the motion characteristics of the robot during palletizing operations can provide critical references for mechanical structure design and key component selection.
2025, 39(1): 7-14.
doi: 10.12299/jsues.24-0033
Abstract:
The ultrasonic vibration assisted drilling process of Al7075-T6/Ti6Al4V under varying rotational speeds and vibration trajectories was analyzed. The drilling process was simulated using the finite element software Deform, with detailed analysis conducted on the workpiece's drilling force, heat and delamination under vibration assisted conditions. The simulation results show that in the thermo-mechanical coupling model of titanium alloy and aluminum alloy drilling, the most significant changes in drilling force and temperature occur when comparing conventional drilling (CD), axial vibration assisted drilling (UAD), and longitudinal torsional vibration assisted drilling (EUAD) model. In terms of delamination, aluminum alloy is more prone to the thermal softening effect than titanium alloy, and the possibility of uplift delamination is greater. The bottom of titanium alloy loses support, and the possibility of pushing away delamination is also high. Ultrasonic vibration changes the force distribution of each layer, reducing the drilling force, temperature and delamination defects during the drilling process. It reveals that EUAD exhibits performance at low rotational speeds, achieving over 30% reduction in axial force, torque and temperature exceeds , whereas its effectiveness decreases by more than 20% at high speed.
The ultrasonic vibration assisted drilling process of Al7075-T6/Ti6Al4V under varying rotational speeds and vibration trajectories was analyzed. The drilling process was simulated using the finite element software Deform, with detailed analysis conducted on the workpiece's drilling force, heat and delamination under vibration assisted conditions. The simulation results show that in the thermo-mechanical coupling model of titanium alloy and aluminum alloy drilling, the most significant changes in drilling force and temperature occur when comparing conventional drilling (CD), axial vibration assisted drilling (UAD), and longitudinal torsional vibration assisted drilling (EUAD) model. In terms of delamination, aluminum alloy is more prone to the thermal softening effect than titanium alloy, and the possibility of uplift delamination is greater. The bottom of titanium alloy loses support, and the possibility of pushing away delamination is also high. Ultrasonic vibration changes the force distribution of each layer, reducing the drilling force, temperature and delamination defects during the drilling process. It reveals that EUAD exhibits performance at low rotational speeds, achieving over 30% reduction in axial force, torque and temperature exceeds , whereas its effectiveness decreases by more than 20% at high speed.
2025, 39(1): 15-20, 72.
doi: 10.12299/jsues.23-0250
Abstract:
The flexible control of collaborative robots relies on the torque sensing capability of the joints. Based on the measurement principle of strain sensors, a strain-based torque sensor with an S-shaped double-hole structure was proposed. Firstly, according to the basic principles of sensor measurement, a configuration design method based on strain gauges was proposed. Secondly, the reliability of the structure was verified through mechanical simulation tests of the overall sensor structure. Finally, the central-composite design (CCD) response surface method was ued to optimize the local dimensional parameters of the sensor, which increases the surface strain of the sensor by 93 MPa, thereby enhancing the sensitivity of the sensor.
The flexible control of collaborative robots relies on the torque sensing capability of the joints. Based on the measurement principle of strain sensors, a strain-based torque sensor with an S-shaped double-hole structure was proposed. Firstly, according to the basic principles of sensor measurement, a configuration design method based on strain gauges was proposed. Secondly, the reliability of the structure was verified through mechanical simulation tests of the overall sensor structure. Finally, the central-composite design (CCD) response surface method was ued to optimize the local dimensional parameters of the sensor, which increases the surface strain of the sensor by 93 MPa, thereby enhancing the sensitivity of the sensor.
2025, 39(1): 21-27, 36.
doi: 10.12299/jsues.24-0023
Abstract:
Compared with integer-order systems, fractional-order systems can better characterize atypical and nonlinear dynamic behaviors, and making them more suitable for describing complex control systems with memory. The observer-based positive consensus problem of fractional-order multiagent systems (FOMAS) was investigated. First, a Luenberger observer was employed to estimate the systems state. By leveraging positive system theory, a sufficient and necessary condition for the positive consensus of FOMAS was obtained. To reduce conservatism, this condition was further optimized through the bounds of improved Laplacian matrix eigenvalues. A less conservative sufficient condition of Riccati inequality type was then established, incorporating the number of nodes in the directed network. Subsequently, by solving the algebraic Riccati inequality, the positive consensus semidfinite programming algorithm for the observer-based FOMAS was presented. Finally, the validity of the obtained results was verified by numerical simulation.
Compared with integer-order systems, fractional-order systems can better characterize atypical and nonlinear dynamic behaviors, and making them more suitable for describing complex control systems with memory. The observer-based positive consensus problem of fractional-order multiagent systems (FOMAS) was investigated. First, a Luenberger observer was employed to estimate the systems state. By leveraging positive system theory, a sufficient and necessary condition for the positive consensus of FOMAS was obtained. To reduce conservatism, this condition was further optimized through the bounds of improved Laplacian matrix eigenvalues. A less conservative sufficient condition of Riccati inequality type was then established, incorporating the number of nodes in the directed network. Subsequently, by solving the algebraic Riccati inequality, the positive consensus semidfinite programming algorithm for the observer-based FOMAS was presented. Finally, the validity of the obtained results was verified by numerical simulation.
2025, 39(1): 28-36.
doi: 10.12299/jsues.24-0034
Abstract:
Aiming at the problem of unbalanced mid-point potential in T-type three-level inverters, the principle of unbalanced mid-point potential in traditional T-type neutral point clamped (NPC) three-level inverters was investigated. To address the mid-point potential imbalance caused by voltage variation on DC side, an optimized topology with an active mid-point potential balancing strategy and a fuzzy PI control were proposed. A simulation system model of the optimized topology T-type three-level inverter was established by Matlab/Simulink, simulation analyses with different control strategies were conducted, and a grid-connected experimental platform was set up to validate the proposed methodology. The simulation and experimental results demonstrate that the fuzzy PI control under the optimized topology can effectively suppress the mid-point voltage fluctuations, with the advantages of fast dynamic response and high response precision, enhancing the disturbance resistance and stability of the inverter system.
Aiming at the problem of unbalanced mid-point potential in T-type three-level inverters, the principle of unbalanced mid-point potential in traditional T-type neutral point clamped (NPC) three-level inverters was investigated. To address the mid-point potential imbalance caused by voltage variation on DC side, an optimized topology with an active mid-point potential balancing strategy and a fuzzy PI control were proposed. A simulation system model of the optimized topology T-type three-level inverter was established by Matlab/Simulink, simulation analyses with different control strategies were conducted, and a grid-connected experimental platform was set up to validate the proposed methodology. The simulation and experimental results demonstrate that the fuzzy PI control under the optimized topology can effectively suppress the mid-point voltage fluctuations, with the advantages of fast dynamic response and high response precision, enhancing the disturbance resistance and stability of the inverter system.
2025, 39(1): 37-43, 85.
doi: 10.12299/jsues.24-0026
Abstract:
A supernumerary robotic finger for disabled people grasping was proposed. The kinematic analysis of the supernumerary finger was carried out, providing mapping relationships between the bending unit and the overall supernumerary finger across the driving space, joint space and operational space. On the basis of a quasi-static analysis of the grasp system, five indexes for measuring the grasp quality were listed. Using the Syngrasp toolbox based on Matlab platform, a grasp model of augmented hand which contained human hand wearing a supernumerary finger was established, grasp simulations of the natural hand and the augmented hand were performed, and the grasp quality indexes were compared. The result demonstrates that wearing the supernumerary finger can improve the quality of grasping objects with the human hand.
A supernumerary robotic finger for disabled people grasping was proposed. The kinematic analysis of the supernumerary finger was carried out, providing mapping relationships between the bending unit and the overall supernumerary finger across the driving space, joint space and operational space. On the basis of a quasi-static analysis of the grasp system, five indexes for measuring the grasp quality were listed. Using the Syngrasp toolbox based on Matlab platform, a grasp model of augmented hand which contained human hand wearing a supernumerary finger was established, grasp simulations of the natural hand and the augmented hand were performed, and the grasp quality indexes were compared. The result demonstrates that wearing the supernumerary finger can improve the quality of grasping objects with the human hand.
2025, 39(1): 44-51.
doi: 10.12299/jsues.24-0037
Abstract:
Multi-brandshared battery swap stations for electric vehicle have become an effective solution to address issues such as low utilization of battery swapping resources and high operational costs. A simulation framework was proposed to model the entire process of electric vehicle station selection, queuing, battery swapping, and departure behaviors, with a focus on investigating how battery quantities of different brands and battery-sharing models affect service quality and operational costs. Based on the location information and operational data of NIO's battery swapping network in Shanghai, multi-scenario simulations were conducted to quantify the impact of different battery configuration schemes on the queuing time for battery swapping users and to calculate the total number of batteries required to maintain a reasonable queuing time. The simulation results demonstrate that by optimizing the number of batteries in the swapping station or adopting a multi-brand battery sharing approach, the service quality of the swapping station can be significantly improved, and the cost of batteries can be reduced.
Multi-brandshared battery swap stations for electric vehicle have become an effective solution to address issues such as low utilization of battery swapping resources and high operational costs. A simulation framework was proposed to model the entire process of electric vehicle station selection, queuing, battery swapping, and departure behaviors, with a focus on investigating how battery quantities of different brands and battery-sharing models affect service quality and operational costs. Based on the location information and operational data of NIO's battery swapping network in Shanghai, multi-scenario simulations were conducted to quantify the impact of different battery configuration schemes on the queuing time for battery swapping users and to calculate the total number of batteries required to maintain a reasonable queuing time. The simulation results demonstrate that by optimizing the number of batteries in the swapping station or adopting a multi-brand battery sharing approach, the service quality of the swapping station can be significantly improved, and the cost of batteries can be reduced.
Feature extraction method of vibration energy of rail vehicle running gear based on CVMD and FastICA
2025, 39(1): 52-57.
doi: 10.12299/jsues.23-0263
Abstract:
To address the processing of nonlinear signals under interference, a vibration energy and feature extraction method integrating correlation coefficient-based variational modal decomposition (CVMD) and fast independent component analysis (FastICA) was proposed, which effectively handles both stationary and non-stationary signals. Through simulations and field test analyses of running gear vibration, it is verified that the main energy distribution of rail vehicle running gear is distributed within 50~120 Hz, and the key characteristic frequencies are identified as 75 and 100 Hz. The finding provide a theoretical basis for vibration energy recovery and feature extraction in the actual engineering applications of metro trains.
To address the processing of nonlinear signals under interference, a vibration energy and feature extraction method integrating correlation coefficient-based variational modal decomposition (CVMD) and fast independent component analysis (FastICA) was proposed, which effectively handles both stationary and non-stationary signals. Through simulations and field test analyses of running gear vibration, it is verified that the main energy distribution of rail vehicle running gear is distributed within 50~120 Hz, and the key characteristic frequencies are identified as 75 and 100 Hz. The finding provide a theoretical basis for vibration energy recovery and feature extraction in the actual engineering applications of metro trains.
2025, 39(1): 58-64.
doi: 10.12299/jsues.23-0260
Abstract:
Occluded person re-identification (ReID) in practical applications faces the main challenges of incomplete or noisy pedestrian features during the extraction and matching processes, which necessitates models with higher stability. Deep learning techniques were employed to construct the ReID model, which incorporates two modules: region attention selection and feature enhancement. The former adaptively selects regions of interest from the original image, while the latter distinguishes human features and applies weighting during the matching process. The ablation analysis demonstrates that these two modules are embeddable and beneficial to occluded pedestrian re-identification task. It can achieve Rank-1 accuracy rates of 55.9% and 79.1% on the Occluded-Duke and Occluded-ReID datasets, respectively.
Occluded person re-identification (ReID) in practical applications faces the main challenges of incomplete or noisy pedestrian features during the extraction and matching processes, which necessitates models with higher stability. Deep learning techniques were employed to construct the ReID model, which incorporates two modules: region attention selection and feature enhancement. The former adaptively selects regions of interest from the original image, while the latter distinguishes human features and applies weighting during the matching process. The ablation analysis demonstrates that these two modules are embeddable and beneficial to occluded pedestrian re-identification task. It can achieve Rank-1 accuracy rates of 55.9% and 79.1% on the Occluded-Duke and Occluded-ReID datasets, respectively.
2025, 39(1): 65-72.
doi: 10.12299/jsues.24-0024
Abstract:
Excavation dewatering is an essential part of foundation pit engineering, but it carries many risks. To ensure the safety of the foundation pit excavation dewatering, a comprehensive risk assessment model was established based on neutrosophic enhanced best-worst method (NE-BWM) and technique for order preference by similarity to ideal solution (TOPSIS). Taking an excavation dewatering of foundation pit in Shanghai as a case study, a risk indicator system was constructed to validate the comprehensive risk assessment model, and sensitivity analysis was performed on the evaluation results. The findings show that this method can yield relatively objective and reliable evaluation results, which are consistent with the actual engineering situation.
Excavation dewatering is an essential part of foundation pit engineering, but it carries many risks. To ensure the safety of the foundation pit excavation dewatering, a comprehensive risk assessment model was established based on neutrosophic enhanced best-worst method (NE-BWM) and technique for order preference by similarity to ideal solution (TOPSIS). Taking an excavation dewatering of foundation pit in Shanghai as a case study, a risk indicator system was constructed to validate the comprehensive risk assessment model, and sensitivity analysis was performed on the evaluation results. The findings show that this method can yield relatively objective and reliable evaluation results, which are consistent with the actual engineering situation.
2025, 39(1): 73-78.
doi: 10.12299/jsues.24-0038
Abstract:
The thermodynamic analysis of the methane dry reforming process was conducted based on Gibbs free energy minimization principle using HSC Chemistry software. The effects of key parameters including reaction temperature, pressure, feed ratio (n(CH4)/n(CO2) mole ratio), O2 and H2O feed rates on equilibrium concentration of CO and H2, n(H2)/n(CO) mole ratio and carbon deposition were systematically investigated. It is shows that under conditions of1137 K reaction temperature, 101.325 kPa pressure, n(CH4)/n(CO2) ratio of 1.20, n(O2)/n(CH4) ratio of 0.15, and n(H2O)/n(CH4) ratio of 0.07, the n(H2)/n(CO) ratio is maintained at approximately 0.90, closely approaching the theoretical value of 1. The methane dry reforming enables both greenhouse gas reduction and high-value utilization, with the produced syngas serving as fuel and important chemical feed gas, making it a promising technology for achieving dual carbon goals.
The thermodynamic analysis of the methane dry reforming process was conducted based on Gibbs free energy minimization principle using HSC Chemistry software. The effects of key parameters including reaction temperature, pressure, feed ratio (n(CH4)/n(CO2) mole ratio), O2 and H2O feed rates on equilibrium concentration of CO and H2, n(H2)/n(CO) mole ratio and carbon deposition were systematically investigated. It is shows that under conditions of
2025, 39(1): 79-85.
doi: 10.12299/jsues.23-0232
Abstract:
To verify the efficacy of Baofukang suppository against Candida tropicalis in vivo and in vitro, a total of 16 clinical strains of Candida tropicalis were collected. The minimum inhibitory concentration (MIC) of Baofukang suppository against Candida tropicalis was determined using microbroth dilution method. The impact on biofilm formation was evaluated by crystal violet staining method, and the alternations in fungus structure was observed using scanning electron microscope (SEM). The vaginal Candida tropicalis infection model in mice was established, changes in vaginal fungus load and vaginal pathology after giving Baofukang suppository were monitored. The results show that Baofukang suppository can inhibit the proliferation and biofilm formation of Candida tropicalis, destroy the cell wall and membrane of Candida tropicalis, effectively relieve vaginal inflammation in mice. Baofukang suppository had obvious inhibitory effect on the growth of Candida tropicalis in vitro and ameliorative effect on vaginitis caused by Candida tropicalis infection in vivo.
To verify the efficacy of Baofukang suppository against Candida tropicalis in vivo and in vitro, a total of 16 clinical strains of Candida tropicalis were collected. The minimum inhibitory concentration (MIC) of Baofukang suppository against Candida tropicalis was determined using microbroth dilution method. The impact on biofilm formation was evaluated by crystal violet staining method, and the alternations in fungus structure was observed using scanning electron microscope (SEM). The vaginal Candida tropicalis infection model in mice was established, changes in vaginal fungus load and vaginal pathology after giving Baofukang suppository were monitored. The results show that Baofukang suppository can inhibit the proliferation and biofilm formation of Candida tropicalis, destroy the cell wall and membrane of Candida tropicalis, effectively relieve vaginal inflammation in mice. Baofukang suppository had obvious inhibitory effect on the growth of Candida tropicalis in vitro and ameliorative effect on vaginitis caused by Candida tropicalis infection in vivo.
2025, 39(1): 86-92.
doi: 10.12299/jsues.24-0039
Abstract:
Long short-term memory (LSTM) networks are widely used for stock price prediction; however, their performance tends to be suboptimal when applied to stock price data with significant fluctuations. A novel WT-L2-LSTM model was proposed, which integrates wavelet transform with L2-LSTM. The original broadband data was decomposed into multiple layers of narrowband data using wavelet transform, and then the L2-LSTM model was used for the prediction layer by layer. The decomposed data is smoother and allows for parameter tuning layer by layer. The prediction results demonstrate that the WT-L2-LSTM model is superior to the LSTM model in terms of goodness of fit and mean square error, indicating that the WT-L2-LSTM model has better accuracy and generalization performance in predicting stock prices.
Long short-term memory (LSTM) networks are widely used for stock price prediction; however, their performance tends to be suboptimal when applied to stock price data with significant fluctuations. A novel WT-L2-LSTM model was proposed, which integrates wavelet transform with L2-LSTM. The original broadband data was decomposed into multiple layers of narrowband data using wavelet transform, and then the L2-LSTM model was used for the prediction layer by layer. The decomposed data is smoother and allows for parameter tuning layer by layer. The prediction results demonstrate that the WT-L2-LSTM model is superior to the LSTM model in terms of goodness of fit and mean square error, indicating that the WT-L2-LSTM model has better accuracy and generalization performance in predicting stock prices.
2025, 39(1): 93-98.
doi: 10.12299/jsues.23-0211
Abstract:
The adaptive finite-time stochastic synchronization problem of a class of neutral-type neural networks was investigated. The neutral-type neural networks with uncertainty and Markovian jumping parameters was constructed. By employing an adaptive control strategy, the finite time stability criterion of the master-slave system was derived. Furthermore, based on the Itô formula and Lyapunov stability theory, sufficient conditions for finite-time stochastic synchronization of neutral-type neural networks were established, and their synchronization time was estimated. The feasibility of the proposed method was demonstrated through numerical examples.
The adaptive finite-time stochastic synchronization problem of a class of neutral-type neural networks was investigated. The neutral-type neural networks with uncertainty and Markovian jumping parameters was constructed. By employing an adaptive control strategy, the finite time stability criterion of the master-slave system was derived. Furthermore, based on the Itô formula and Lyapunov stability theory, sufficient conditions for finite-time stochastic synchronization of neutral-type neural networks were established, and their synchronization time was estimated. The feasibility of the proposed method was demonstrated through numerical examples.
2025, 39(1): 99-105.
doi: 10.12299/jsues.24-0030
Abstract:
The addition of aluminum oxide (Al2O3) nanoparticle to minimal quantity lubrication (MQL) base oils forms nanofluid minimal quantity lubrication (NMQL). Using NMQL machining can further enhance the tribological state and improve the surface quality of workpiece. A three-dimensional computational fluid dynamic (CFD) model was developed and simulated to investigate the flow field and heat transfer mechanism during NMQL machining of Ti6Al4V. Through the CFD model, a parametric cross-analysis of different flow rates and inlet pressures was conducted to explore the effects of NMQL on cutting heat under various cooling conditions and the selection principles of optimal process parameters. Finite element analysis of turbulent two-phase flow and heat transfer was performed to sumrize the effect of adding Al2O3 nanoparticle to MQL base oil on the temperature distribution across the tool surface. The results show that the average tool temperature decreases with increasing Al2O3 nanoparticle concentration, and the maximum temperature near the tool tip decreases by about 6% on average, thus effectively improving the temperature distribution in the cutting area. The exploration of the flow field and heat transfer during NMQL machining of Ti6Al4V can provide valuable references for studying machining modes and mechanisms of difficult-to-process materials such as Ti6Al4V, offering both theoretical significance and application value.
The addition of aluminum oxide (Al2O3) nanoparticle to minimal quantity lubrication (MQL) base oils forms nanofluid minimal quantity lubrication (NMQL). Using NMQL machining can further enhance the tribological state and improve the surface quality of workpiece. A three-dimensional computational fluid dynamic (CFD) model was developed and simulated to investigate the flow field and heat transfer mechanism during NMQL machining of Ti6Al4V. Through the CFD model, a parametric cross-analysis of different flow rates and inlet pressures was conducted to explore the effects of NMQL on cutting heat under various cooling conditions and the selection principles of optimal process parameters. Finite element analysis of turbulent two-phase flow and heat transfer was performed to sumrize the effect of adding Al2O3 nanoparticle to MQL base oil on the temperature distribution across the tool surface. The results show that the average tool temperature decreases with increasing Al2O3 nanoparticle concentration, and the maximum temperature near the tool tip decreases by about 6% on average, thus effectively improving the temperature distribution in the cutting area. The exploration of the flow field and heat transfer during NMQL machining of Ti6Al4V can provide valuable references for studying machining modes and mechanisms of difficult-to-process materials such as Ti6Al4V, offering both theoretical significance and application value.
2025, 39(1): 106-112.
doi: 10.12299/jsues.24-0025
Abstract:
In response to the challenge that current hardware performance and network bandwidth are insufficient to support real-time rendering of massive data in 3D scenes, a web-based building information modeling (BIM) scene roaming solution was proposed. Initially, the BIM model was reconstructed, and a top-down hierarchical structure was employed to achieve fine-grained instances of scene components. Subsequently, model compression was completed based on the spatial position of vertices and the similarity calculation between components to shorten the display delay of the scene. Finally, a visual component picking algorithm based on the two-layer architecture of view frustum ball and view frustum was designed to complete scene selection and rendering by off-line component number and reduce hardware load. Five different scale scenarios were selected for validation. The results demonstrate an average compression rate of 45%, which enables smooth 30 frames per second roaming for GB-scale scene in web-based applications.
In response to the challenge that current hardware performance and network bandwidth are insufficient to support real-time rendering of massive data in 3D scenes, a web-based building information modeling (BIM) scene roaming solution was proposed. Initially, the BIM model was reconstructed, and a top-down hierarchical structure was employed to achieve fine-grained instances of scene components. Subsequently, model compression was completed based on the spatial position of vertices and the similarity calculation between components to shorten the display delay of the scene. Finally, a visual component picking algorithm based on the two-layer architecture of view frustum ball and view frustum was designed to complete scene selection and rendering by off-line component number and reduce hardware load. Five different scale scenarios were selected for validation. The results demonstrate an average compression rate of 45%, which enables smooth 30 frames per second roaming for GB-scale scene in web-based applications.
2025, 39(1): 113-118.
doi: 10.12299/jsues.23-0206
Abstract:
To enhance the accuracy of text classification models and address the deficiencies of traditional word vector models in syntax, semantics, and deep-level information representation, a text classification model based on the ERNIE 3.0 pre-trained model with BiLSTM_Attention was proposed. First, the ERNIE 3.0 model was used to encode text dataset, generating word vectors with rich semantic information. Subsequently, text features were extracted through the BiLSTM layer and the Attention layer. Finally, the output word vectors were classified via the Softmax layer. Classification experiments conducted on the THUCNews dataset compared the accuracy and F1-score metrics across different models. The results show that the ERNIE 3.0_BiLSTM_Attention model achieves superior classification performance.
To enhance the accuracy of text classification models and address the deficiencies of traditional word vector models in syntax, semantics, and deep-level information representation, a text classification model based on the ERNIE 3.0 pre-trained model with BiLSTM_Attention was proposed. First, the ERNIE 3.0 model was used to encode text dataset, generating word vectors with rich semantic information. Subsequently, text features were extracted through the BiLSTM layer and the Attention layer. Finally, the output word vectors were classified via the Softmax layer. Classification experiments conducted on the THUCNews dataset compared the accuracy and F1-score metrics across different models. The results show that the ERNIE 3.0_BiLSTM_Attention model achieves superior classification performance.
2017, 31(2): 174-177,182.
doi: 10.3969/j.issn.1009-444X.2017.02.016
摘要：
随着纺织印染行业的不断发展,纺织印染废水成分也日益复杂,大大增加了废水处理难度.针对纺织印染废水的来源和水质特点,综述了国内外纺织印染废水处理技术的研究进展,包括物理处理法、化学处理法和生物处理法,分析了各类处理方法的机制、处理效果及其优缺点.对纺织印染废水处理技术的发展前景进行了展望,指出优化组合废水处理工艺可提高纺织印染废水处理效果及综合回用率.
随着纺织印染行业的不断发展,纺织印染废水成分也日益复杂,大大增加了废水处理难度.针对纺织印染废水的来源和水质特点,综述了国内外纺织印染废水处理技术的研究进展,包括物理处理法、化学处理法和生物处理法,分析了各类处理方法的机制、处理效果及其优缺点.对纺织印染废水处理技术的发展前景进行了展望,指出优化组合废水处理工艺可提高纺织印染废水处理效果及综合回用率.
2018, 32(1): 64-67.
doi: 10.3969/j.issn.1009-444X.2018.01.014
摘要：
交通数据质量是影响交通控制方法有效实施的关键因素之一.为进一步提升交通数据的准确性和时效性,对比分析交通预处理方法,采用标准差公式对拉依达准则进行优化,设计一种适用于交通异常数据检测和修复的优化方法,并结合上海城市快速路实际数据对模型有效性和时效性进行检验.结果表明,基于拉依达准则的数据处理优化方法能够有效、实时地检测交通异常数据、改善数据质量,为道路交通状况的监测和预警提供数据支持.
交通数据质量是影响交通控制方法有效实施的关键因素之一.为进一步提升交通数据的准确性和时效性,对比分析交通预处理方法,采用标准差公式对拉依达准则进行优化,设计一种适用于交通异常数据检测和修复的优化方法,并结合上海城市快速路实际数据对模型有效性和时效性进行检验.结果表明,基于拉依达准则的数据处理优化方法能够有效、实时地检测交通异常数据、改善数据质量,为道路交通状况的监测和预警提供数据支持.
2020, 34(3): 238-246.
doi: 10.3969/j.issn.1009-444X.2020.03.006
摘要：
针对机械臂关节空间轨迹规划的时间优化问题,结合机械臂运动约束,提出基于非线性动态改变惯性权重的粒子群优化(NPSO)算法.根据传统3-5-3多项式插值方法,采用改进粒子群算法寻求最短关节运动时间,研究线性递减改变惯性权重(LPSO)算法和NPSO算法的性能,选用NPSO算法完成关节运动时间最优求解.研究结果显示,经时间优化后的3-5-3插值曲线连续光滑且具备更好的运动特性,整体运动时间缩短约26%,证实提出的方法具有可行性.
针对机械臂关节空间轨迹规划的时间优化问题,结合机械臂运动约束,提出基于非线性动态改变惯性权重的粒子群优化(NPSO)算法.根据传统3-5-3多项式插值方法,采用改进粒子群算法寻求最短关节运动时间,研究线性递减改变惯性权重(LPSO)算法和NPSO算法的性能,选用NPSO算法完成关节运动时间最优求解.研究结果显示,经时间优化后的3-5-3插值曲线连续光滑且具备更好的运动特性,整体运动时间缩短约26%,证实提出的方法具有可行性.
2017, 31(4): 371-375.
doi: 10.3969/j.issn.1009-444X.2017.04.017
摘要：
老年人参与各项社会活动,能够延缓衰老,促进身体健康、提高生活质量.通过开展调研探究了老年人的社会参与意愿及其影响因素.调查发现,城市老年人对经济、政治、社区、文化等各项活动的参与意愿都较高,但是多数的参与停留在被动层面;年龄、性别、文化程度、健康状况等因素会影响参与意愿,且不同活动受影响的程度不同;参与意愿会受到现实条件的约束,如照看小孩,料理家务等.
老年人参与各项社会活动,能够延缓衰老,促进身体健康、提高生活质量.通过开展调研探究了老年人的社会参与意愿及其影响因素.调查发现,城市老年人对经济、政治、社区、文化等各项活动的参与意愿都较高,但是多数的参与停留在被动层面;年龄、性别、文化程度、健康状况等因素会影响参与意愿,且不同活动受影响的程度不同;参与意愿会受到现实条件的约束,如照看小孩,料理家务等.
2021, 35(3): 237-242.
摘要：
针对传统光流法处理视频序列时存在运行效率低及跟踪偏移问题，结合粒子滤波模型提出一种改进光流法的视频目标跟踪技术. 该技术首先通过遍历法搜索运动点，采用质心定位方式捕获目标质心坐标，然后将得到的视频序列进行光流处理，最后经粒子滤波求解质心运动信息，以实现对视频中目标的精确检测与追踪. 在不同场景下对不同数量、不同类型目标进行仿真跟踪试验，并与光流法、ViBe算法及YOLO算法进行比较. 仿真结果表明，该跟踪技术可使目标跟踪的精准率有效提高5.2%，跟踪效率提高13.7%，同时表现出较好的鲁棒性.
针对传统光流法处理视频序列时存在运行效率低及跟踪偏移问题，结合粒子滤波模型提出一种改进光流法的视频目标跟踪技术. 该技术首先通过遍历法搜索运动点，采用质心定位方式捕获目标质心坐标，然后将得到的视频序列进行光流处理，最后经粒子滤波求解质心运动信息，以实现对视频中目标的精确检测与追踪. 在不同场景下对不同数量、不同类型目标进行仿真跟踪试验，并与光流法、ViBe算法及YOLO算法进行比较. 仿真结果表明，该跟踪技术可使目标跟踪的精准率有效提高5.2%，跟踪效率提高13.7%，同时表现出较好的鲁棒性.
Numerical Simulation Analysis on Air Distribution of Inter-Column Air Conditioning in a Machine Room
2018, 32(3): 244-249.
doi: 10.3969/j.issn.1009-444X.2018.03.009
摘要：
以上海某数据中心机房为研究对象,针对机房服务器散热量大的特点及传统封闭冷通道存在的问题,提出适合该机房的封闭热通道方案.采用计算流体动力学(CFD)模拟的方法,分析机房内封闭热通道与传统封闭冷通道的温度场和速度场分布特性.通过模拟分析得出:大散热量机房列间空调采用封闭热通道形式效果更佳.
以上海某数据中心机房为研究对象,针对机房服务器散热量大的特点及传统封闭冷通道存在的问题,提出适合该机房的封闭热通道方案.采用计算流体动力学(CFD)模拟的方法,分析机房内封闭热通道与传统封闭冷通道的温度场和速度场分布特性.通过模拟分析得出:大散热量机房列间空调采用封闭热通道形式效果更佳.
2018, 32(3): 214-220.
doi: 10.3969/j.issn.1009-444X.2018.03.004
摘要：
介绍功能纺织品的分类,分析功能纺织品的开发途径,主要阐述国内现代功能纺织品研究及开发进展,包括纳米光触媒多功能纺织品、防电磁辐射纺织品、超疏水多功能纺织品、阻燃纺织品、防蚊虫纺织品,并指出采用高新技术开发生态安全高附加值多功能纺织品是功能性纺织品的发展趋势.
介绍功能纺织品的分类,分析功能纺织品的开发途径,主要阐述国内现代功能纺织品研究及开发进展,包括纳米光触媒多功能纺织品、防电磁辐射纺织品、超疏水多功能纺织品、阻燃纺织品、防蚊虫纺织品,并指出采用高新技术开发生态安全高附加值多功能纺织品是功能性纺织品的发展趋势.
2017, 31(1): 90-94.
doi: 10.3969/j.issn.1009-444X.2017.01.019
摘要：
针对我国黄金期货价格预测问题,对影响我国黄金期货价格的5项指标进行灰色关联度分析,得出我国黄金期货价格与美国黄金期货价格之间的关联度最高.建立反向传播(BP)神经网络模型对我国黄金期货价格预测,并与GM(1,1)方法和ARIMA(0,2,1)模型下的预测结果进行对比.结果显示:与后两个模型相比,BP神经网络模型在黄金期货价格预测方面的精确性更高,具有较好的实用价值.
针对我国黄金期货价格预测问题,对影响我国黄金期货价格的5项指标进行灰色关联度分析,得出我国黄金期货价格与美国黄金期货价格之间的关联度最高.建立反向传播(BP)神经网络模型对我国黄金期货价格预测,并与GM(1,1)方法和ARIMA(0,2,1)模型下的预测结果进行对比.结果显示:与后两个模型相比,BP神经网络模型在黄金期货价格预测方面的精确性更高,具有较好的实用价值.
2018, 32(4): 324-328.
doi: 10.3969/j.issn.1009-444X.2018.04.007
摘要：
综述人工神经网络在轧制参数预报中的应用,包括轧制力能参数、组织性能和温度的预报;并介绍了混合人工神经网络模型与改进的人工神经网络模型,其中混合对象包括传统数学模型、有限元和其他智能算法.研究表明,人工神经网络在轧钢参数预报中具备很大的优势,能显著提高轧钢相关参数预报精度,提升轧材的品质.
综述人工神经网络在轧制参数预报中的应用,包括轧制力能参数、组织性能和温度的预报;并介绍了混合人工神经网络模型与改进的人工神经网络模型,其中混合对象包括传统数学模型、有限元和其他智能算法.研究表明,人工神经网络在轧钢参数预报中具备很大的优势,能显著提高轧钢相关参数预报精度,提升轧材的品质.
2019, 33(3): 215-218,277.
doi: 10.3969/j.issn.1009-444X.2019.03.005
摘要：
综述了二氧化硅(SiO2)薄膜的制备方法和研究进展,介绍了磁控溅射、溶胶凝胶法和真空镀等SiO2薄膜制备方法及其优缺点,论述SiO2薄膜在光学、电学和光电等性能研究方面的新进展,最后对SiO2薄膜的应用和发展进行了展望.
综述了二氧化硅(SiO2)薄膜的制备方法和研究进展,介绍了磁控溅射、溶胶凝胶法和真空镀等SiO2薄膜制备方法及其优缺点,论述SiO2薄膜在光学、电学和光电等性能研究方面的新进展,最后对SiO2薄膜的应用和发展进行了展望.
2003, 17(2): 83-86.
doi: 10.3969/j.issn.1009-444X.2003.02.001
Abstract:
2019, 33(3): 283-284,288.
doi: 10.3969/j.issn.1009-444X.2019.03.017
Abstract:
About Journal
More>
Chinese Quarterly
Supervisor: Shanghai Municipal Education Commission
Sponsors by: Shanghai University of Engineering Science
Publisher: Editorial Department of Journal of Shanghai University of Engineering Science
Editor-in-Chief: WANG Yansong
Email: xuebao@sues.edu.cn
ISSN1009-444X
CN31-1598/T
Link
More>
Email Alert
RSS
