上海工程技术大学学报

2024, 38(3): 1-2.

Abstract:

[Abstract](65) [FullText HTML] (39) [PDF 368KB](13)

Study on high frequency radial electromagnetic wave of motor driven by frequency conversion

WANG Zhen, GUO Hui, FENG Qingbao, WANG Yansong, LIU Ningning

2024, 38(3): 231-239. doi: 10.12299/jsues.23-0199

Abstract:
By using the field-route coupling method, the causes and main components of the high frequency radial electromagnetic force characteristic frequency of permanent magnet synchronous motor were studied. The generation mechanism and harmonic characteristics of harmonic current were analyzed by analytical method, and verified by numerical simulation. The characteristic frequency distribution of high frequency radial electromagnetic force was analyzed by means of the magnetomotive force permeability method and Maxwell tensor method. The air gap magnetic field obtained by finite element calculation was decomposed in time and space, and the high frequency radial electromagnetic force generated by magnetic field interaction was calculated by Maxwell tensor method. The results show that the high frequency radial electromagnetic force of PMSM driven by frequency conversion is mainly generated by the interaction between the harmonic magnetic field and the stator fundamental magnetic field introduced by the harmonic current of the frequency converter, which has certain guiding significance for the research of PMSM.

[Abstract](105) [FullText HTML] (47) [PDF 1396KB](25)

Numerical simulation and experimental study of ultrasonic machining of aluminum honeycomb core

WU Nan, ZHANG Liqiang, YANG Jie, LU Licheng, LIU Gang

2024, 38(3): 240-249. doi: 10.12299/jsues.23-0190

Abstract:
Traditional machining techniques often result in damage to the aluminum honeycomb. In order to improve the surface quality of aluminum honeycomb machining, the numerical and simulation models of cutting force in ultrasonic-assisted machining of aluminum honeycomb were established. The influence of mesh size on simulation results was investigated to determine an appropriate mesh size. Furthermore the cutting forces and surface morphology were analyzed for different machining parameters. After a thorough evaluation, a mesh of 0.2 mm was chosen for the simulation. The results show that there is a clear reduction in machining damage with increasing the feed rate, spindle speed, and cutting width values. This conclusion can provide reference for mathematical model, simulation and experiment of ultrasonic-assisted machining of aluminum honeycomb.

[Abstract](67) [FullText HTML] (28) [PDF 5837KB](11)

Intelligent wheelchair obstacle avoidance design based on fuzzy PID control

WANG Yu, WANG Dazhi, ZHOU Yufeng

2024, 38(3): 250-256. doi: 10.12299/jsues.23-0216

Abstract:
In order to improve obstacle avoidance accuracy of intelligent wheelchair, a control algorithm of intelligent wheelchair based on fuzzy PID control was proposed. The kinematics model of intelligent wheelchair was established, and the intelligent wheelchair control system based on Fuzzy PID control was designed on the basis of traditional PID control method. Matlab was used to construct simulation tests, the fuzzy controller was designed and constructed in Simulink, and the obstacle avoidance performances of traditional and fuzzy PID control were simulated. The experimental results show that the control algorithm can optimize the obstacle avoidance error of intelligent wheelchair, and it has advantages of small overshoot, fast response, and higher accuracy compared with the traditional PID control algorithm.

[Abstract](113) [FullText HTML] (30) [PDF 1647KB](6)

Estimation method of train boarding and alighting time in urban rail transit station

HE Yuxin, HU Hua, PU Xuexiang, HAO Yanxi, FANG Yong

2024, 38(3): 257-263. doi: 10.12299/jsues.23-0212

Abstract:
Passenger boarding and alighting time is an important part of the dwelling time and operation diagram of urban rail transit trains, and is also a vital link that affect the operation efficiency and safety of rail transit lines. In this paper, the passenger boarding and alighting process was divided into two independent processes. Based on survey data, the influencing factors of the maximum passenger flow and the average boarding and alighting time were analyzed. Additionally, estimation submodel of the maximum passenger flow and the average boarding and alighting time of passengers were established. On this basis, the estimation model of train boarding and alighting time for rail transit stations was constructed. Taking Shanghai rail transit station as an example, the analysis showed that the deviation between the measured values of boarding and alighting time and the estimated value of the model is within 11.89%, which verified the effectiveness and practicability of the model. The research results have important practical value for improving the matching degree between the actual and planned boarding and alighting time of rail transit trains, enhancing train stopping efficiency, and ensuring passenger boarding and alighting safety.

[Abstract](75) [FullText HTML] (23) [PDF 1120KB](16)

Research on target recognition algorithm and control strategy of AEB system in curved roads

WU Jianhui, WU Changshui

2024, 38(3): 264-271. doi: 10.12299/jsues.23-0210

Abstract:
A target recognition method based on curve coordinate transformation was proposed to address the challenge of misidentification in existing automatic emergency braking (AEB) systems under curved conditions. The geometric information of the road model could be reflects through sensors, the curve coordinate transformation method was used to locate the position of the main vehicle and the target vehicle, the relative distance between vehicles were calculated, and the dangerous target vehicle was determined by using the logic gate restriction method. A longitudinal collision avoidance control strategy combining the Honda algorithm and TTC algorithm were proposed to address the traditional collision avoidance algorithms based on the safe distance model or time-to-collision (TTC) algorithm, which cannot balance the safety and comfort issues during the braking process. Using the TTC algorithm as a forward collision warning strategy, and the autonomous emergency braking strategy was designed based on the optimized Honda algorithm. The simulation results show that the proposed method based on curve coordinate transformation can accurately calculate the distance between the main vehicle and the target vehicle, accurately and efficiently identify dangerous target vehicles. The fusion algorithm based on the collaborative control of the safety distance algorithm and TTC algorithm can effectively avoid longitudinal car following collisions, and take into account the safety and comfort of the emergency braking process.

[Abstract](100) [FullText HTML] (32) [PDF 2059KB](4)

Design of Unity3D-based virtual simulation training platform for trackside signaling equipment

ZHANG Jie, LIU Zhigang, WEI Lili

2024, 38(3): 272-277. doi: 10.12299/jsues.23-0195

Abstract:
A virtual simulation training platform based on Unity3D trackside signaling equipment was developed, and three main functional modules of three-dimensional display, animation demonstration and interactive evaluation were designed. Firstly, Autodesk Revit was used to model the trackside signaling equipment in detail, and the C# scripting language was used in the Unity3D engine to control the three-dimensional display of the model. Then, the open-source animation component Animator Controller was used to create a multi-scene mission demonstration animation. Finally, the 3D model was spliced and assembled by using the collision detection principle and the mouse drag event, and the equipment replacement process was evaluated. Research result shows that the training platform can overcome problems of low efficiency poor sense of space of two-dimensional drawings of traditional teaching and training, enhance the sense of immersion, meet the needs of maintenance personnel for knowledge training and simulation training, which provides a new means for trackside signal equipment training.

[Abstract](54) [FullText HTML] (26) [PDF 1524KB](3)

Comparison of railway passenger flow forecast methods based on ARIMA and LSTM

YU Yanqiao, LI Sijie, LIU Zhigang

2024, 38(3): 278-283. doi: 10.12299/jsues.23-0236

Abstract:
Accurate passenger flow forcast is the basis of the optimization of passenger transportation organization and an effective way to improve operational safety and transportation efficiency. Taking the annual passenger flow data of Jiangmen East Railway Station as research object, and the ARIMA time series model and the LSTM neural network model were constructed respectively. From the perspectives of forecast accuracy, calculation speed, error index evaluation, and model adaptability, the differences in passenger flow forecast results of two forecast models were analyzed. The results show that the LSTM model has better forecast accuracy and fitting accuracy, and the ARIMA model has faster calculation speed. This study result can provide significant practical implications for the selection of passenger flow forecast methods.

[Abstract](113) [FullText HTML] (45) [PDF 843KB](11)

Wind power prediction method based on IWOA-ELM

PU Shibiao, ZENG Guohui, LIU Jin

2024, 38(3): 284-290. doi: 10.12299/jsues.23-0191

Abstract:
In wind energy storage microgrids, accurately predicting the actual output power of wind farms in advance can effectively improve the stability of grid integration regulation. To address the low prediction accuracy of existing models for wind power characteristic parameters, a wind power characteristic parameter prediction method based on improved whale optimization algorithm of extreme learning machine (IWOA-ELM) was proposed. By optimizing the parameters of the extreme learning machine using an improved whale optimization algorithm, a IWOA-ELM wind power characteristic parameter prediction model based on time series was established to predict the characteristic parameters of future wind power. Model performance was evaluated using metrics such as root mean square error and mean absolute error. Experimental results show that the proposed prediction method has a root mean square error of 5.488% and a mean absolute error of 3.72% for wind speed prediction, and a root mean square error of 19.354% and a mean absolute error of 12.46% for wind direction prediction. The prediction accuracy is significantly higher than that of other wind power prediction models such as WOA-ELM, PSO-ELM, BP, and ELM.

[Abstract](46) [FullText HTML] (30) [PDF 1115KB](3)

Analysis of pumping test and optimization design of excavation dewatering in complex geological environment

PAN Kai, WU Yongxia, LU Jiansheng

2024, 38(3): 291-297. doi: 10.12299/jsues.23-0189

Abstract:
Aiming at the dewatering of deep excavation pit in complex and thick confined aquifers, relied on an engineering case in Shanghai, the measured data of pumping test before excavation were verified by numerical simulation, the hydrogeological parameters of each aquifer were obtained through the inversion and verification of the measured data. The ground settlement caused by dewatering with the preliminary designed diaphragm wall was predicted by the verified model, and the comparison of groundwater drawdown and ground settlement under four different diaphragm wall depths was analyzed. The results indicate that ground settlement caused by pumping is large with the existed design of the diaphragm wall depths. The greater the depth of the diaphragm wall, the smaller the ground settlement caused by dewatering. When the diaphragm wall depth is greater than the pumping well depth, the variation of the ground settlement caused by dewatering is small. Considering the environmental effects and the diaphragm wall cost, the optimal depth determined is 48.8 m for the diaphragm wall.

[Abstract](57) [FullText HTML] (31) [PDF 2204KB](1)

Design of indoor thermal comfort system based on wireless sensor network

WEN Jian, DENG Shengxiang

2024, 38(3): 298-303. doi: 10.12299/jsues.23-0194

Abstract:
In order to minimize energy consumption while maintaining thermal comfort, an indoor thermal comfort system based on wireless sensor network and simplified predicted mean vote (PMV) index was designed. The scatter layout method was used to determine the best measurement nodes of indoor sensors, and the collected data was transmitted to the system through ZigBee communication. In the system, the thermal comfort perception of indoor environment was evaluated by simplified PMV index. Finally, fuzzy control with fuzziness and non-linearity was used to regulate the operation of the air conditioner. The results show that the data collected by the scatter sensor is more stable and accurate. Simplified PMV index can effectively replace the PMV index in evaluating human thermal comfort perception. The thermal comfort system can not only control the indoor environment within the thermal comfort range, but also play a good energy-saving effect.

[Abstract](45) [FullText HTML] (45) [PDF 1190KB](4)

Measurement and influencing factors study of spatial quality in urban waterfront areas along the Suzhou River in Shanghai supported by multi-source data

ZHAN Jiajie

2024, 38(3): 304-312, 327. doi: 10.12299/jsues.23-0193

Abstract:
The spatial quality of urban waterfront areas is an important indicator of the level of urban development and construction. Traditional research on spatial quality has problems such as limited spatial scale, strong subjectivity, and single research perspective. Based on Open Street Map road data, Baidu Map street views, and Baidu Map point-of-interest (POI) data, combing with machine learning technology, the spatial quality of waterfront area along the Suzhou River in Shanghai was measured. MGWR2.2.1 software was used to establish a multi-scale geographically weighted regression model for influencing factors of spatial quality. The result shows that the western Putuo District, western Changning District, and the north bank of Suzhou River in Jingan District have lowspatial quality in the study area, which should be prioritized during optimization. The eastern part of the study area is suitable for improving spatial quality through measures such as planting trees, building pedestrian paths, and opening pedestrian streets, while the western part should focus on developing leisure and consumer industries such as catering, shopping, and entertainment, as well as conjunct them with existing park green space resources, more trees should be planted, pocket parks are also necessary. The study results can provide a reference for optimizing the spatial quality of the Suzhou River waterfront area.

[Abstract](51) [FullText HTML] (27) [PDF 1655KB](2)

Offline reinforcement learning dynamic obstacles avoidance navigation algorithm

GE Qixing, ZHANG Wei, XIE Guiliang, HU Zhi

2024, 38(3): 313-320. doi: 10.12299/jsues.23-0227

Abstract:
Real-time sampling and updating of data to optimize obstacle avoidance strategies for unmanned aerial vehicle (UAV) is an urgent issue in applying deep reinforcement learning (DRL) to collision prevention. In response to this problem, a dynamic obstacle avoidance navigation algorithm based on offline DRL was proposed. The combination of an offline DRL algorithm and velocity obstacle (VO) algorithm was introduced to address the issue of the high real-time interaction data required by online DRL algorithms. Performance enhancement of the offline DRL algorithm was achieved by imposing constraints on policy updates. A reward function based on VO was developed, which could enable the UAV to consider both time consumption and the shortest path while avoiding dynamic obstacles. Simulation verification in a three-dimensional obstacle navigation environment show that this method can surpass online deep reinforcement learning obstacle avoidance algorithms in terms of path length, flight time, and obstacle avoidance success rate. It will effectively address the problem of DRL requiring continuous input of online data for efficient policy updates

[Abstract](76) [FullText HTML] (35) [PDF 1556KB](2)

Positive edge consensus of multiagent systems on directed graphs

ZHANG Pengyu, ZHANG Wei, HU Zhi

2024, 38(3): 321-327. doi: 10.12299/jsues.23-0197

Abstract:
Most of the existing literature on the problem of positive edge consensus of multi-agent systems has mainly focused on undirected graphs or strongly connected directed graphs. To extend it to the directed networks containing spanning trees, since the Laplacian matrix of a directed network containing spanning trees may be complex, its analysis may become very difficult. Using positive system theory and graph theory, the necessary and sufficient conditions for edge system to achieve positive consensus under a directed network containing spanning trees were given. The results were further optimized by improving the bounds on the eigenvalues of the Laplace matrix, sufficient conditions involving only the number of edge number of the nodal network were obtained. Riccati inequality was solved and a semidefinite programming algorithm was developed to obtain the solution. Finally, the validity of the obtained results was verified by numerical simulation.

[Abstract](55) [FullText HTML] (25) [PDF 1015KB](4)

Integrated cross-efficiency model based on directional distance function and information entropy

NING Yangxue, PAN Haojun, WANG Guoqiang

2024, 38(3): 328-333. doi: 10.12299/jsues.23-0022

Abstract:
Aiming at the problem that traditional cross-efficiency model cannot handle both input and output data containing negative numbers, an integrated cross-efficiency model based on directional distance function and information entropy was proposesed. First of all, the idea of direction distance function was used to deal with negative numbers. Secondly, the complete ranking of decision units was realized by combining cross efficiency. Then, with the help of the variation coefficient of information entropy, a set of public weights for cross-efficiency integration were obtained to avoid the weight deviation of the traditional model and retain the decision information in the evaluation process. Finally, the effectiveness and practicability of the proposed model were verified by a numerical example, and the research scope and application scenarios of the cross-efficiency model were extended.

[Abstract](55) [FullText HTML] (29) [PDF 673KB](2)

Application research of visual-based method for detecting toilet cleanliness

PENG Yong, ZHANG Hua, GAO Yanfeng

2024, 38(3): 334-340, 348. doi: 10.12299/jsues.23-0209

Abstract:
At present, intelligent nursing robots for secondary defecation only rinse the bucket with a constant amount of water after defecation, without corresponding cleanliness testing. The remaining feces can easily cause air odor and bacterial growth, and even spread diseases. In response to this, two visual based cleanliness detection methods were proposed: one is to evaluate the cleanliness of the toilet by calculating the proportion of dirty pixels, and the other is to use image template matching method to detect the cleanliness of the toilet. Experimental results show that by using adaptive threshold segmentation to calculate the proportion of dirty pixels, it can effectively overcome the influence of background shadows and accurately distinguish photo groups with different levels of pollution, thus achieving the goal of detecting the cleanliness of the toilet.

[Abstract](39) [FullText HTML] (24) [PDF 2277KB](2)

Improved dragonfly algorithm based on chaos mapping and differential evolution

TIAN Bowen, ZHU Zina, LIU Rui, GUO Zhongyang

2024, 38(3): 341-348. doi: 10.12299/jsues.23-0241

Abstract:
An improved dragonfly algorithm based on chaotic mapping and differential evolution was proposed to address several issues encountered in the original algorithm. The initial population’s significant randomness and the difficulty in adjusting algorithmic weight parameters led to low convergence accuracy. Additionally, later-stage population contraction restrictions resulted in decreased vitality and slow convergence speed. By employing Tent chaotic mapping, the population’s initial distribution was made more uniform. The alignment, clustering, and inertia weights were adjusted to enhance convergence speed and accuracy. The introduction of the differential evolution algorithm aimed to accelerate convergence at the final stages. Finally, nine test functions were selected for comparative simulation experiments. The results demonstrated that, compared to the basic dragonfly algorithm and the differential evolution dragonfly algorithm, the improved dragonfly algorithm based on chaotic mapping and differential evolution has significantly improved convergence speed and accuracy, avoiding getting stuck in local optima and obtaining stable and reliable global optimal solutions.

[Abstract](57) [FullText HTML] (27) [PDF 1842KB](4)

Current Issue 2024 Vol. 38, No. 3

Current Issue
2024 Vol. 38, No. 3