中文核心期刊

高校精品期刊Ei收录期刊

Advanced Search

2022 Vol. 42, No. 12

column
Previous Issue| Next Issue
Display Method:
cover
cover
2022, 42(12)
Abstract(71) PDF(28)
Abstract:
2022, 42(12): 1-2.
Abstract(47) HTML(23) PDF(11)
Abstract:
2022, (12): 1328-1328.
Abstract(45) HTML(20) PDF(8)
Abstract:
Engineering Mechanics
Research on Design of Linear Focusing Warhead and Fragmentation Characteristics
GE Chao, WANG Jin, ZHENG Yuanfeng, WANG Haifu, YU Qingbo
2022, 42(12): 1219-1228. doi:10.15918/j.tbit1001-0645.2021.349
Abstract(123) HTML(40) PDF(29)
Abstract:
In order to further enhance the fragment focusing effect of the warhead, this research was done focusing on the design of linear focusing warhead and fragmentation characteristics. Based on linear focusing principle, an arrangement method of fragments was analyzed. A theoretical model for equal-impulse charge generatrix was established by introducing correction parameter of charge length. By numerical modeling, the fragmentation characteristics of linear focusing warhead were obtained, while the effects of charge length and correction parameter on average fragment velocity, velocity amplitude, and standard deviation were revealed. The results show that when charge length is constant, average velocity of fragments increases with the increase of correction parameter, but velocity amplitude and standard deviation of velocity distribution decrease. When correction parameter is constant, average velocity, velocity amplitude and standard deviation of fragments increase with the increase of charge length. The results can provide a new technology approach for the design of fragmentation warhead.
Study on Residual Stress Field Reconstruction of Aluminum Alloy Plate Based on Finite Measurements
LIU Guangyan, XIONG Tulin, WANG Lu, WEN Lei
2022, 42(12): 1229-1235. doi:10.15918/j.tbit1001-0645.2021.361
Abstract(132) HTML(94) PDF(18)
Abstract:
The whole residual stress field of structural components can be reconstructed based on measured results from a limited number of points by the stress function method, but stress function is usually difficult to determine and it is not easy to reconstruct the distribution of several stress components through one stress function. To overcome these shortcomings, a method was developed to simultaneously reconstruct several residual stress component fields by using a general stress function. In this method, the binary Fourier function was used as the Airy stress function. The order of the model was determined by the determination coefficient, and the unknown parameters in the stress function were optimized according to measured results on some points. Finally, the accuracy of this method was verified by simulated residual stress fields of a three-point bending beam, and it was applied to the reconstruction of residual stress field of a real aluminum alloy plate. The results show that this method can simultaneously reconstruct several stress components with high accuracy.
Study on Dynamic Impact Engraving Characteristic of Cased Telescoped Ammunition
CHANG Renjiu, XUE Xiaochun, YU Yonggang
2022, 42(12): 1236-1245. doi:10.15918/j.tbit1001-0645.2022.001
Abstract(93) HTML(42) PDF(13)
Abstract:
To study the high-strain-rate plastic deformation and fracture failure of the projectile belt at the slope during the dynamic impact extrusion process of cased telescoped ammunition, considering the particularity of the secondary ignition of the cased telescoped ammunition and the combustion of the gunpowder program, with the bottom pressure of the projectile and the one-load speed of the projectile obtained from the test taken as the initial boundary conditions of the numerical simulation model, a three-dimensional finite element model was built based on the structural characteristics of the body tube and the cased telescoped ammunition, large deformation and damage effect of the material. Based on the LS-DYNA software, the dynamic engraving process of the cased telescoped ammunition was studied using an explicit numerical calculation method. The variation law of projectile impact extrusion resistance was obtained, and the forming process and stress strain characteristics of the grooves on the surface of the elastic belt were analyzed. The results show that when the copper belt of the cased telescoped ammunition is violently engraved into the slope chamber, it undergoes a process from elastic deformation to plastic deformation and finally to fracture failure. When the elastic belt is completely engraved into the body tube full depth rifling, the entire dynamic impact engraving process is over, and a deep groove is formed on the surface of the elastic belt, which closely adheres to the body tube rifling, during which the impact engraving resistance shows a strong nonlinear change.
Mechanical Engineering
A Damage Prediction Model of Wet Friction Elements Based on PSO-BP Neural Network
LI Le, SHU Yuechao, WU Jianpeng, CHEN Man, WANG Liyong
2022, 42(12): 1246-1255. doi:10.15918/j.tbit1001-0645.2021.347
Abstract(131) HTML(56) PDF(17)
Abstract:
In order to solve the multi factor damage relationship of wet clutch, a wet friction element damage prediction model based on PSO-BP neural network was constructed by using multi-source data fusion method. Taking rotational speed and joint oil pressure as input parameters of the model, taking the extracted circumferential temperature gradient of friction plate, the change rate of Fe and Cu concentration and the change rate of friction plate surface roughness Ra as output parameters of the model, a finite element simulation model was established, and the comprehensive friction and wear test-bed of wet clutch was built. The effects of oil pressure and speed on the damage characteristic parameters of friction elements were studied by using the control variable method. The results show that the input condition takes on a nonlinear relationship with the four types of damage characteristic parameters, the variation trend of the predicted value and the measured value is consistent with the working condition, and the damage characteristic parameters are more sensitive than the change of oil pressure. Compared with similar models and test data, the prediction model can provide higher prediction accuracy and can effectively predict the multi condition damage of wet clutch.
An Eco-Driving Method with Queue Length Estimation for Connected Vehicles
ZHANG Chuntao, LENG Jianghao, WANG Bo, SUN Chao, ZHOU Xingyu
2022, 42(12): 1256-1263. doi:10.15918/j.tbit1001-0645.2021.368
Abstract(129) HTML(70) PDF(21)
Abstract:
Aiming at speed planning problems for connected vehicles traveling through multiple traffic signals under a dynamic traffic environment, an eco-driving method was proposed based on real-time queue length estimation. Firstly, a radial basis function neural network was constructed and trained to estimate queue length at intersection. Then, in the frame of optimal control, the traffic queuing was mathematically modeled together with traffic signals to formulate a speed profile optimization problem. Finally, the proposed decoupling transformation method was used to calculate a reference speed profile efficiently. Simulation results reveal that the proposed method can provide smoother actual speed profiles and save more than 40% energy compared with the traditional eco-driving method without considering the traffic queuing.
Usability Evaluation of In-Car Instant Messaging Applications with Multimodal Interactions
MA Jun, PAN Weitao, XU Wenxia
2022, 42(12): 1264-1272. doi:10.15918/j.tbit1001-0645.2022.002
Abstract(139) HTML(80) PDF(17)
Abstract:
Usability evaluation model was constructed for in-car instant messaging applications with multimodal interactions, and optimization suggestions were proposed based on evaluation results. With research in human-machine interface and usability, usability principles including “clear and easy to operate”, “efficient and practical”, and “safe” were determined, and 15 indexes were defined accordingly. Usability evaluation tests were conducted on three cars by building a real-vehicle driving simulation platform innovatively. Combining AHP and CRITIC method, index weights were determined dynamically by considering subjective opinions and objective data performance, and evaluation model was built by applying fuzzy comprehensive evaluation. Finally, problems such as visual and manual distraction caused by touch-screen interaction negatively affecting safety, and task completion time of voice interaction directly affecting user trust and thus driving performance, etc. were identified, and targeted design suggestions were put forward. The real-vehicle driving simulation platform and evaluation model can be used to test the usability of and evaluate more interaction models, tasks and infotainment applications, and thus guide experience design.
Informatics and Control
3D Target Detection Method Combined with Multi-View Mutual Projection Fusion
ZHAO Yanan, WANG Xiancai, GAO Li, LIU Yujia, DAI Yu
2022, 42(12): 1273-1282. doi:10.15918/j.tbit1001-0645.2021.332
Abstract(118) HTML(64) PDF(24)
Abstract:
Aiming at the lack of feature fusion of multi-sensor target regions in the current target detection of intelligent vehicles, a three-dimensional target detection method was proposed based on multi-modal information fusion. Firstly, taking the image view and aerial view of lidar point cloud as input, the target detection was optimized by an improved AVOD deep learning network algorithm. And then, a multi-angle joint loss function was inducted to prevent the branch network image degradation. Finally, a dual-view image and the lidar point cloud projected mutual fusion method was presented to enhance data spatial correlation and to carry out feature fusion. The experimental results show that the improved AVOD-MPF network can improve the detection accuracy of small-scale targets while retaining the advantages of the AVOD network for vehicle target detection, and achieve 3D target detection with feature-level and decision-level fusion.
Research on Real-Time Scene Classification of Autonomous Platform Based on Randomization-Based Network
DAI Yingpeng, WANG Junzheng, LI Jing
2022, 42(12): 1283-1289. doi:10.15918/j.tbit1001-0645.2022.007
Abstract(80) HTML(29) PDF(8)
Abstract:
Scene classification enables the autonomous platform to understand the environmental information. For the scene classification task, randomization-based neural networks could quickly recognize the scene information and spend little time to train the weights. However, the shallow network structure of randomization-based neural networks limits the non-linear representation ability. Moreover, the fully connected method ineffectively extracts local feature information and introduces a large number of parameters. Ensemble architecture could effectively improve the accuracy. However, it introduces high computational complexity and lots of parameters, which will greatly slow down during inference. To tackle above problems, a multi-level convolutional randomization-based network ensemble architecture (E-MCRNet) was proposed for real-time scene classification tasks. Firstly, replacing the fully connected layer with multi-level convolutional layer, the randomization-based network was constructed a multi-level convolutional randomization-based network (MCRNet). Secondly, multiple MCRNets were combined to form an ensemble architecture named E-MCRNet. The E-MCRNet consists of one main-hidden layer and multiple sub-hidden layers. The main-hidden layer was concatenated with each sub-hidden layer to form component networks respectively. Testing results show that E-MCRNet can improve the accuracy and decrease model complexity. Moreover, it can be deployed on embedded equipment to deal with relevant tasks.
Optics and Electronics
MS-LMB Filter for Stealthy Target Tracking Based on Greedy Measurement Partitioning Mechanism
SUN Jinping, DAI Beining, ZHANG Yutao
2022, 42(12): 1290-1298. doi:10.15918/j.tbit1001-0645.2021.353
Abstract(109) HTML(55) PDF(14)
Abstract:
For stealthy target tracking, we apply the greedy measurement partitioning mechanism to the multi-sensor labeled multi-Bernoulli (MS-LMB) filter to solve the problem of multi-radar tracking under low detection probability. Generally, Gibbs sampling is applied to solve the measurement partitioning problem in the traditional MS-LMB filter. However, when most radars in the radar network are in the state of missing detection due to low detection probability, the likelihood weight of the stealthy target will be too small to be easily obtained by Gibbs sampling. Thus, it’s difficult to accurately estimate the state of stealthy target. The greedy measurement mechanism can solve this problem since it separately considers the measurement set containing missing detection items. The simulation results show that the filtering performance of the MS-LMB filter with greedy measurement partitioning mechanism is obviously better than that of the MS-LMB filter with Gibbs sampling when tracking stealthy targets.
A Multi-Bit 3D RRAM-Based Signed Floating-Point Number Operations
WANG Xinghua, WANG Tian, WANG Qian, LI Xiaoran
2022, 42(12): 1299-1304. doi:10.15918/j.tbit1001-0645.2021.358
Abstract(167) HTML(52) PDF(13)
Abstract:
In this paper, a signed floating-point number operation with multi-bit storage three-dimensional resistive random-access memory (3D RRAM) was presented for complex convolution neutral network (CNN) systems. Comparing with other types of memory, 3D RRAM can not only perform calculations inside the memory, but also possess a higher reading rate and a lower energy consumption, providing a new solution to the bottleneck problem of the Von Neumann architecture. A single RRAM cell can reach a maximum and minimum resistance of 10 GΩ and 10 MΩ, which can be stabilized in multi-level resistance states to store high-bit-width data. The test results show that, the accuracy of the signed floating-point number convolution operation system can reach up to 99.8%, the measured peak reading speed of the 3D RRAM model is 0.529 MHz.
Robust Beamforming of Polarization Array Based on an Improved General Linear Combination Algorithm
LÜ Yan, CAO Fei, YANG Jian, FENG Xiaowei, HE Chuan
2022, 42(12): 1305-1311. doi:10.15918/j.tbit1001-0645.2021.362
Abstract(109) HTML(65) PDF(18)
Abstract:
To improve the robustness of beamforming for polarization array, a general linear combination (GLC) algorithm was applied to the polarization array in this paper. Firstly, analyzing the reason of output signal to interference plus noise ratio decrease with the increase of the number of snapshots under the conditions of sensor disturbance and direction of arrival (DOA) mismatch of the signal of interest (SOI) at high input signal to noise ratio, an improved GLC algorithm was proposed. And then, the input signal to noise ratio was identified by the proposed algorithm according to the size of parameters related to the eigenvalues of the sample covariance matrix (SCM). At high input signal to noise ratio, the GLC algorithm combined with the conversion function was used to calculate the diagonal loading level (DLL). And at a low level of the input signal to noise ratio, the original GLC algorithm was used to calculate the DLL, making the output signal to interference plus noise ratio of the improved GLC algorithm be greater than or equal to that of the original GLC algorithm under any input signal to noise ratio and snapshot. Finally, the effectiveness of the proposed algorithm was verified by simulations under the condition of mainlobe interference.
Efficient Hardware Acceleration System Design for End-to-End Object Detection Neural Network
REN Shiwei, LIU Chaojia, LI Jianzheng, JIANG Rongkun, WANG Xiaohua, XUE Chengbo
2022, 42(12): 1312-1320. doi:10.15918/j.tbit1001-0645.2022.004
Abstract(101) HTML(48) PDF(19)
Abstract:
To solve the problem of limited hardware resources and sensitive power consumption in the application of neural network object detection system for edge computing devices, a YOLOv3-Tiny neural network object detection hardware acceleration system was proposed based on field programmable gate array (FPGA). The scale of YOLOv3-Tiny network was reduced by using network structure reorganization, inter layer fusion and dynamic numerical quantization. Based on channel parallel and weight resident hardware acceleration algorithm, tight pipeline processing flow and hardware operation unit reuse, the utilization efficiency of hardware resources was improved. The designed end-to-end object detection acceleration system was deployed on UltraScale+ XCZU9EG FPGA. The result shows that it can achieve 96.6 GOPS throughput, 17.3 FPS detection frame rate and 4.12 W power consumption. The hardware resource utilization efficiency is 0.32 GOPS/DSP and 2.68 GOPS/kLUT. Maintaining efficient and accurate object detection capability, the utilization efficiency of hardware resources is better than other existing YOLOv3-Tiny object detection hardware accelerators.
Chemical Engineering and Materials Science
Study on the Application of Composite Conductive Agent in Ultrafine Hollow Carbon Spheres Supercapacitors
ZHAO Yun, GUO Yiqing, ZHAO Xiaohuan, LIANG Jie, JIAO Qingze
2022, 42(12): 1321-1328. doi:10.15918/j.tbit1001-0645.2022.017
Abstract(90) HTML(39) PDF(6)
Abstract:
Whether the conductive agent could form a good conductive network in the electrode materials is one of the key factors that affect the performance of supercapacitor. Ultrafine hollow carbon spheres with high specific surface area and hierarchical pore structure, synthesized based on the modified Stöber method, were taken as the electrode materials of supercapacitors. And the effects of single-walled carbon nanotubes/carbon black composite conductive agents on the performance of supercapacitor were studied with single carbon black or single-walled carbon nanotubes as the contrast. The results show that at current density of 0.2 A/g, the specific capacitance of sample with the composite conductive agents is 205.7 F/g, being much higher than that of carbon black or carbon nanotubes alone. At a high current density of 100 A/g, the specific capacitance can remain a high value of 104.0 F/g for the composite conductive agent sample, heightening 275% than that of carbon black. The examining results indicate that the long fibrous carbon nanotubes and carbon black particles can form a dot-line synergistic conductive network in the ultrafine hollow carbon spheres, which is the main reason of improving the performance of supercapacitors.

Copyright：Academic Journal Office of Beijing Institute of Technology

Phone： 010-68914629Email：journal@bit.edu.cnAddress: 5 Zhongguancun South st, Haidian District, Beijing

Supported by:Beijing Renhe Information Technology Co., Ltd.

Baidu
map