上海工程技术大学学报

2024, 38(4): 1-2.

Abstract:

[Abstract](24) [FullText HTML] (16) [PDF 394KB](1)

Study on effect of diesel injection strategy on performance of ammonia-diesel dual-fuel engine

CHEN Xu, MIAO Xuelong, ZHENG Jinbao, DI Yage, ZHAO Zhifeng, GUO Lixin

2024, 38(4): 355-362. doi: 10.12299/jsues.23-0271

Abstract:
Based on a heavy-duty diesel engine, the effects of different ammonia energy ratio (AER) and diesel injection strategies on the combustion performance of ammonia-diesel dual-fuel engine were investigated by using CONVERGE software, the indicated thermal efficiency (ITE) of the engine was optimized by adjusting the diesel injection strategy. The results of the study show that the AER increases from 0% to 80% with a consequent decrease in the ITE, the increase in the reaction product N₂O resulted in greater greenhouse gas (GHG) emissions than in the original diesel-only combustion mode of the diesel engine. The AER is constant at 50%. At the optimum point of diesel single injection timing, the ITE is comparable to that of the original diesel combustion mode, while GHG emissions decreases by 19.7% and NO emissions increases by 41.6%. At the optimum point of pre- and main injection timing for diesel split injection, the ITE increases by 1.6%, GHG emission decreases by 40.4%, and NO emission increased by 31.8% compared with the original diesel combustion mode, delaying the main injection timing of diesel split injection can effectively reduce NO emissions.

[Abstract](46) [FullText HTML] (18) [PDF 1128KB](6)

Improved discrete differential evolution algorithm for solving vehicle routing problem

YU Kaiying, XU Bin

2024, 38(4): 363-369. doi: 10.12299/jsues.23-0215

Abstract:
Given that the vehicle routing problem with capacity constraints is easily influenced by uncertain factors such as customer location and demand, and leads to the infeasible or non-optimal solution, an improved discrete differential evolution algorithm was proposed. The greedy method was employed to construct the initial solution, thereby enhancing its quality of the initial solution. The mutation operator was redesigned according to the integer permutation characteristics, and the crossover operator was adapted for problem model, which improved the algorithm performance effectively. A local search strategy that incorporates local relocation, conditional exchange and a maximum contribution removal mechanism were designed, enhancing algorithm's exploration capabilities. Simulation experiments on the benchmark test set show that the improved algorithm can find the optimal value for 31 out of 33 test cases, indicating superior solution capabilities compared to other algorithms and effectively addressing the vehicle routing problem with capacity constraints.

[Abstract](20) [FullText HTML] (14) [PDF 834KB](0)

Guiding strategy of stair and escalator passengers in metro hub stations considering urban resilient safety

DING Xiaobing, HONG Chen, SHI Gan

2024, 38(4): 370-374. doi: 10.12299/jsues.24-0010

Abstract:
To address issues related to passenger congestion at the stairs and escalators of the platforms resulting from the intense gathering of passenger flow upon the arrival of rail transit trains, the impedance function of the optimal decision selecting stair and escalator was constructed, which integrated both physical consumption and time consumption. Firstly, the number of passengers disembarking from each door of the arriving train were collected and Dijkstra's algorithm was utilized to allocate passengers to stair and escalator groups. Secondly, passenger queuing parameters at stairs and escalators were calculated. Thirdly, combined with the queuing characteristics, the impedance functions for stair and escalator selection were built, then the optimal decision of stair and escalator selection for different passengers were obtained. Finally, a case study of Hongqiao Railway Station of Shanghai Metro Line 2 was carried out. The results indicated that, when the waiting time in the queue before the escalator was 10.32 s, it was recommended for regular passengers to use the stairs; when the waiting time in the queue before the escalator increased to 29.91 and 40.31 s, it was advisable for passengers with luggage to use the stairs.

[Abstract](34) [FullText HTML] (12) [PDF 794KB](1)

Identification of key cognitive ability factors for metro train drivers based on VTS data mining

SHI Zhanwang, YANG Jufen, ZHU Haiyan

2024, 38(4): 375-381. doi: 10.12299/jsues.24-0001

Abstract:
Cognitive abilities of 354 metro train drivers were assessed by using the Vienna test system (VTS). An unsupervised learning model was developed through K-means clustering algorithm on the VTS data to establish a cognitive ability classification model. With the maximum Recall value as the objective function, XGBoost training and optimization were performed on the classification model. SHAP algorithm was employed to analyze the importance of various cognitive ability feature indicators in the model, and three key factors that mean reaction time, total correct responses, visual field range, as well as their interactions were identified. The research results can provide a more precise tool for the selection, on-the-job assessment, and training of metro train drivers when applied to the field of cognition and emergency capabilities.

[Abstract](26) [FullText HTML] (15) [PDF 1240KB](2)

Impact of bipolar plate flow channel cross-section shape on PEMFC mass transfer and performance

HAO Zhaoyang, ZHENG Jinbao, MIAO Xuelong, DI Yage

2024, 38(4): 382-388. doi: 10.12299/jsues.23-0218

Abstract:
The performance of parallel single-flow channel fuel cells with a rectangular cross-section was investigated using COMSOL software. A performance simulation model was established and its feasibility was verified. On this basis, the impacts of different cross-sectional shapes on the power density, mass transfer, temperature, velocity, pressure drop, and net output efficiency of parallel single channel fuel cells were examined. The result shows that the cross-sectional shape of the flow channel significantly affects the fuel cell performance under low voltage and high current density conditions. The triangular cross-section model generally exhibits a higher maximum power density. The W-shaped cross-section has a higher flow velocity and smaller temperature change compared to the rectangular cross-section, with an increase of 1.8% in hydrogen consumption and a 6.6% in oxygen consumption respectively. It demonstrates a higher power density with a similar net output efficiency, providing support for the design of bipolar plates.

[Abstract](22) [FullText HTML] (11) [PDF 1377KB](0)

Experimental study on frosting characteristics of vertical fins

XU Conghui, XIA Peng

2024, 38(4): 389-394. doi: 10.12299/jsues.23-0201

Abstract:
In winter, the surface of the vertical fin of the cold end of the outdoor air intake device is prone to frost. Through visualization experiments, the characteristics of frost formation on the surface of vertical fins under different air conditions were studied to provide a basis for suppressing frost formation on the surface of fins. The results show that the frost crystal morphology of fin surface is affected by the cold surface temperature, and the frost crystal show irregular shape, fan shape, feather shape and tree shape in turn when the cold surface temperature decreases. When set the relative humidity of the air to 60% and increase the air temperature within the range of 10 to 25 ℃, it was found that the freezing time of water droplets on the surface of the fin was shortened from 273 to 205 s and then extended to 269 s. The average frost growth rate was first fast and then slow. When the air temperature is setted to 20 ℃ and the relative humidity of air increases within the range of 20% ~ 80%, the freezing time of water droplets on the surface of fin is gradually shortened from 345 to 218 s, and the average growth rate of frost layer is faster. The moisture content of the air is approximately equal, the air temperature increases within the range of 10 to 25 ℃, and the average growth rate of the frost layer slows down.

[Abstract](20) [FullText HTML] (11) [PDF 1472KB](0)

Friction and wear properties of graphene reinforced titanium matrix composites

ZHANG Jingwen, GE Zhangyi, WAN Zhaomei, LI Jiuxiao, MA Xiaopei, ZHANG Rui, WANG Renjie, ZHENG Yan, TAO Junzhou

2024, 38(4): 395-399. doi: 10.12299/jsues.23-0272

Abstract:
Titanium alloy has low density, high specific strength, high stiffness, high temperature resistance and corrosion resistance. However, its low hardness, poor thermal conductivity, high viscosity and poor wear resistance have become an important problem restricting the application of titanium alloys. Graphene reinforced titanium matrix composites were prepared by powder metallurgy, the effects of graphene on microstructure, hardness, tensile properties and friction and wear properties were studied. The results show that the grain size of titanium matrix composites is smaller than that of matrix, and the grain size decreases with the increase of graphene content. Compared with the matrix, the hardness, strength and friction and wear properties of titanium matrix composites are significantly improved. With the increasing of graphene content, the hardness and strength increase, the plasticity decreases, and the friction coefficient and wear amount both decrease. When the sintering temperature is 1273 K and the mass fraction of graphene is 0.5%, the hardness, tensile and friction wear properties of titanium matrix composites are the best, and the wear loss is the lowest, which is 40% lower than that of matrix.

[Abstract](30) [FullText HTML] (15) [PDF 1484KB](0)

Analysis of impact of manufacturing defects on bearing capacity of carbon fiber aluminum honeycomb sandwich structures

SHEN Qin, LU Qiang, LI Junli, LIU Gang

2024, 38(4): 400-405. doi: 10.12299/jsues.23-0229

Abstract:
To investigate reasons for the impact of manufacturing defects on bearing performance of carbon fiber aluminum honeycomb sandwich structures, flat compression tests and three-point bending tests under quasi-static conditions were conducted on these structures with glued joints, the different forms of debonding failure existing in the same batch and size were analyzed. The trends of peak load and average load generated by the sandwich structure were observed and compared with the load-displacement curves of non-debonded specimens simulated using Abaqus CAE software, to study the changes in its bearing capacity. The results show that sandwich structures with manufacturing defects exhibit different forms of debonding and have lower bearing capacity than non-debonded specimens, which can provide a basis for improving the fabrication process and have great significance for broadening its application ares.

[Abstract](29) [FullText HTML] (17) [PDF 2314KB](0)

Rolling bearing fault diagnosis method based on adaptive spectral loss generative adversarial networks

HE Jiaxing, CHEN Xingjie, LYU Zhaomin

2024, 38(4): 406-413. doi: 10.12299/jsues.23-0254

Abstract:
A rolling bearing fault diagnosis method based on adaptive spectrum loss generative adversarial networks was proposed. Firstly, the spectral distance was introduced to measure the frequency domain difference between generated data and real data. Secondly, adaptive spectrum loss was added to the generator loss to reduce the weight of simple frequency components, so as toadaptively focus on difficult-to-synthesize frequency components, to better guide the generative adversarial networks to generate fake samples more similar to real data. The proposed method was validated using the Case Western Reserve University bearing dataset. Compared with other methods, the adaptive spectrum loss generative adversarial networks can generate higher-quality samples and significantly improve fault recognition rate under sample imbalance conditions.

[Abstract](27) [FullText HTML] (13) [PDF 1427KB](0)

Short-term wind prediction based on mRMR and SVMD-TPA-BiSTM

XING Peiyu, WEI Yunbing, HU Hua, ZHANG Wenhu

2024, 38(4): 414-421. doi: 10.12299/jsues.23-0217

Abstract:
A short-term wind power prediction method based on mRMR and SVMD-TPA-BiLSTM was proposed. Firstly, successive variational modal decomposition (SVMD) was adopted to reduce the dimensionality of wind power sequences, and the decomposed subsequences were combined with the key meteorological feature data selected maximum correlation minimum redundancy (mRMR) to form a training set. Secondly, a combined model based on the temporal pattern attention (TPA) mechanism to improve the Bi-directional long short-term memory neural network (BiLSTM) was established, and TPA mechanism was used to capture the correlation of different time sequence data. Finally, the prediction component results were superimposed to obtain the final prediction results. Taking the dataset of a certain wind farm as an example, different models were subjected to single step and multi-step prediction experiments to compare the accuracy, and it was proved that the prediction method can effectively portray the component characteristics of wind power and improve the prediction accuracy of wind power.

[Abstract](45) [FullText HTML] (14) [PDF 1684KB](1)

Kinematic modeling and prototype test of flexible supernumerary robotic finger in grasping for the disabled

SUN Weijun, XU Yong, GUO Shuyan, LIU Lingxiao

2024, 38(4): 422-428. doi: 10.12299/jsues.23-0262

Abstract:
A wearable flexible supernumerary robotic finger system capable of collaborative grasping with disabled arms was proposed, and its structure and prototype were designed. A kinematic model for supernumerary robotic finger bending and grasping objects was established, the forward and inverse kinematic mapping relationship between the stretching displacement of the driving rope, the joint angle of the finger joint, and the pose of the end finger joint were obtained. The dynamics of the flexible supernumerary robotic fingers were carried out through simulation, the movement rules of the unloaded finger end were obtained. Through experimental construction, the flexible supernumerary robotic finger prototype can achieve stable envelope grasping of objects 1.7 times larger than the body. It verifies the rationality of kinematic modeling and structural design, and has certain practicality.

[Abstract](24) [FullText HTML] (13) [PDF 2042KB](0)

Attitude PID control parameter tuning of curtain wall cleaning robot based on improved genetic algorithm

GAO Bo, WU Minghui, HU Heping, YANG Chen

2024, 38(4): 429-436, 464. doi: 10.12299/jsues.24-0007

Abstract:
To address the problem of time-consuming and large errors in the attitude PID parameter tuning of a high-rise curtain wall cleaning robot, an improved genetic algorithm (IGA) was proposed. The halton sequence was introduced as the initial population, the adaptive dynamic regulation mechanism was introduced in the crossover and variation stages, and improved integral of time-weighted absolute error (IITAE) evaluation index function with better comprehensive performance was proposed. PID parameter tuning experiments were carried out using IGA, manual empirical method, AGPSO and GA. The results show that the tuning results of the IGA algorithm are more than 20% higher than other algorithms in terms of the objective function value and reduced the convergence time by more than 50%. The controller designed by the IGA methodcan achieve stable control of robot attitude, which has good application value for air attitude stability control of curtain wall cleaning robot.

[Abstract](23) [FullText HTML] (15) [PDF 2340KB](1)

Research and application of fuzzy PID precise control for robot moxibustion temperature

LYU Xianji, XIONG Genliang, LI Zeguang, PU Zeyang, YU Chao, LI Yinxin

2024, 38(4): 437-443. doi: 10.12299/jsues.23-0245

Abstract:
In order to accurately control the surface skin temperature of moxibustion and improve the therapeutic effect of moxibustion, an intelligent moxibustion temperature control system was designed based on a moxibustion robot. The fuzzy PID control algorithm was used to stably control the surface skin temperature of moxibustion. Simulation result shows that the system temperature error always fluctuates within ± 0.4 ℃, and the maximum overshoot is less than 1.3 ℃. Compared with the conventional PID control system, the intelligent moxibustion temperature control system based on the fuzzy PID has smaller temperature overshoot, faster stabilization time, better temperature control accuracy, and can improve patient comfort.

[Abstract](27) [FullText HTML] (11) [PDF 1644KB](1)

Segmentation of lung tumors based on PCU-Net

CAI Hao, LI Peng, GONG Xiaomei, WANG Raofen

2024, 38(4): 444-450. doi: 10.12299/jsues.24-0012

Abstract:
Deep learning techniques can assist doctors in precise tumor segmentation. However, existing methods often suffer from issues such as fuzzy segmentation edges and large model parameter counts due to the unclear boundaries between lung tumors and surrounding tissues. A partial convolution coordinate attention U-net (PCU-Net) algorithm for lightweight lung tumor segmentation was proposed. The partial convolution was introduced to reduce model parameters and enhance feature extraction capability. The coordinate attention module was added at skip connection of PCU-Net, so that more precise localization of tumors was achieved by network and segmentation accuracy was improved. The research result shows that the improved PCU-Net can reduce model parameters by 58.57% while increase Dice coefficient, Intersection over Union (IoU) and Recall by 4.22%, 4.26% and 6.82%, respectively. The comparison between PPU-Net and other semantic segmentation models shows that Dice coefficient of PCU-Net is 3-6 percentage points higher than that of other models.

[Abstract](19) [FullText HTML] (15) [PDF 2663KB](0)

High-quality dynamic real-time rendering method based on conditional generative adversarial networks

JIANG Likai, WANG Guozhong, ZHAO Haiwu

2024, 38(4): 451-457. doi: 10.12299/jsues.24-0015

Abstract:
Focusing on the challenge of real-time rendering in computer graphics, integrating rasterization techniques with optimized conditional generative adversarial networks (CGANs), real-time generation of approximate ray-traced images was achieved, the issue of discontinuity between frames in existing research was effectively addressed, and optimized balances among real-time performance, realism, and visual coherence were achieved. Based on Pix2PixGAN architecture, the structure, data input and loss functions of CGANs were improved, a training rendering dataset using by Unity and Blender was constructed. Experimental results demonstrate that our rendering method can surpass traditional approaches in key performance metrics, enhance the quality of image generation and the coherence between frames.

[Abstract](21) [FullText HTML] (12) [PDF 1708KB](0)

Analysis and prediction of temperature distribution of steel trusses for kilometer-scale rail-cum-road bridge

ZHENG Gankang, HE Yuelei, MENG Xiaoliang, WAN Leshan

2024, 38(4): 458-464. doi: 10.12299/jsues.24-0018

Abstract:
Based on the temperature observation of a kilometer-scale rail-cum-road bridge, temperature distribution characteristics of main girder steel trusses at different locations in different seasons were investigated, and the temporal and spatial temperature field characteristics of the kilometer-scale rail-cum-road bridge were obtained. Based on variation characteristics of the structure and atmospheric temperature, a comprehensive analysis of temperature field of a kilometer-scale rail-cum-road bridge was conducted. The most suitable model for accurate prediction of bridge temperature data was found by comparing three neural network prediction models. The results show that the temperature field of steel trusses have significant difference in time and space; the temperature of main girder rises more slowly than atmospheric temperature; LSTM and CNN neural networks can predict the temperature data with high accuracy.

[Abstract](22) [FullText HTML] (11) [PDF 1470KB](1)

Heart disease identification based on boosted decision tree

TANG Jiayao, YU Su

2024, 38(4): 465-470. doi: 10.12299/jsues.24-0219

Abstract:
A gradient boosting decision tree (BDTG) model based on high-energy physics data analysis in ROOT framework was proposed for the identification of heart disease using a multivariate analysis method. Through a large amount of clinical data, the aim is to analyze the various complex relationships of variables to improve the accuracy and reliability of heart disease differentiation. Using the Kaggle open-source heart disease dataset, the results showed that the model did not exhibit any erroneous discrimination when the BDTG responsevalues range between between −0.4 and 0.5. In addition, when the truncation of BDTG response values is −0.6 or 0.6, the model still maintained above 98% in accuracy, precision, recall and F1 scores. Therefore, the model has high accuracy and reliability in the diagnosis of heart disease. This study not only provides new perspectives and methods for predicting heart disease, but also serves as a reference for machine learning prediction research on other diseases.

[Abstract](25) [FullText HTML] (17) [PDF 811KB](2)

Current Issue 2024 Vol. 38, No. 4

Current Issue
2024 Vol. 38, No. 4