Loading...

Archive

    For Selected: Toggle Thumbnails
    Information & Control
    Automatic Lane Change Decision Model Based on Dueling Double Deep Q-network
    ZHANG Xue-feng, WANG Zhao-yi
    2023, 44 (10):  1369-1376.  DOI: 10.12068/j.issn.1005-3026.2023.10.001
    Abstract ( 986 )   HTML ( 96)   PDF (2568KB) ( 598 )  
    Automatic lane change of vehicles requires driving at the fastest possible speed while ensuring no collision situations. However, regular control is not robust enough to handle unexpected situations or respond to lane separation. To solve these problems, an automatic lane change decision model based on dueling double deep Q-network(D3QN) reinforcement learning model is proposed. The algorithm processes the environmental vehicle information fed back by the internet of vehicles, and then obtains actions through strategies. After the actions are executed, the neural network is trained according to given reward function, and finally the automatic lane change strategy is realized through the trained network and reinforcement learning. The three-lane environment built by Python and the vehicle simulation software CarMaker are used to carry out simulation experiments. The results show that the algorithm proposed has a good control effect, making it feasible and effective.
    References | Related Articles | Metrics
    Research on Calibration Adaptation Method via Variational Inference for Near-Infrared Spectroscopy
    ZHAO Yu-hui, QI Tian-shu, LU Peng-cheng
    2023, 44 (10):  1377-1383.  DOI: 10.12068/j.issn.1005-3026.2023.10.002
    Abstract ( 440 )   HTML ( 30)   PDF (1156KB) ( 151 )  
    In near-infrared spectroscopy analysis, existing calibration transfer methods are mostly based on standard samples and non-parametric induction models, which generally suffer from short model lifespan, limited model applicability. To address this problem, a variational inference calibration adaptation(VICA)method is proposed, which aligns the feature distributions of the source domain(master instrument)and a target domain(slave instrument)by a parametric method. VICA performs principal component analysis on the source domain data and establishes a variational regression model for the source domain features. During prediction, VICA first projects the target domain data into the source domain feature subspace, and then establishes a distribution difference function between the source and target domain features, and obtains the probability density model of the target domain by minimizing this function, achieving model transfer. Experimental comparison results show that VICA performs better in calibration transfer than most existing methods.
    References | Related Articles | Metrics
    Online Content Caching and Delivery Method Based on Edge Horizontal Collaboration
    LIU Ming-han, CHEN Xiang-yi, CHEN Xue-ping, ZHAO Hai
    2023, 44 (10):  1383-1391.  DOI: 10.12068/j.issn.1005-3026.2023.10.003
    Abstract ( 516 )   HTML ( 19)   PDF (1136KB) ( 151 )  
    Traditional network architectures cannot meet user needs for content caching, and there is a conflict between low latency requirements and high communication costs in content delivery. To address these problems, an online content caching and delivery algorithm based on Lyapunov optimization and branch and bound method is proposed under the scenario of horizontal collaboration between edge nodes to balance delivery delay and cost and to make efficient decisions on content caching and content delivery. The proposed algorithm decomposes the continuous problem into a single slot online optimization problem based on Lyapunov optimization theory, and solves them using branch and bound algorithms. Simulation experiments showed that the proposed algorithm can achieve lower average content delivery delay and higher content hit rate under a limited content delivery cost budget. It can also adaptively balance content delivery delay and delivery cost.
    References | Related Articles | Metrics
    Self-adaptation Adjusting Window Width and Window Level Algorithm for Medical CT Sequence Images
    CHEN Jin-lin, YUAN Pei-xin, HOU Hao-nan, ZHAO Zhao
    2023, 44 (10):  1392-1400.  DOI: 10.12068/j.issn.1005-3026.2023.10.004
    Abstract ( 513 )   HTML ( 21)   PDF (3667KB) ( 423 )  
    Window transformation can improve the accuracy of gray level recognition in medical CT images. The default window(127.5, 255)has strong limitations. Manual window transformation requires rich priori knowledge and is inefficient, making it unsuitable for widespread use. Therefore, a self-adaptation algorithm for window transformation in medical CT sequence images is proposed. Firstly, a histogram is drawn according to the maximum and minimum gray level values of the sequence images. Secondly, relevant parameters are set to traverse the histogram to remove parts with frequencies below the threshold T0. After traversing the histogram again, parts with a difference between adjacent groups of frequencies less than the threshold T1 are merged. Finally, the window width and window level are calculated based on the histogram. Experimental results showed that the algorithm improved the mean square error, signal-to-noise ratio and peak signal-to-noise ratio, which could effectively improve the accuracy of gray level recognition.
    References | Related Articles | Metrics
    DOA Estimation Algorithm of Mixed Signals Based on Oblique Projection Operator
    SHE Li-huang, ZHANG Jian-yu, ZHANG Shi
    2023, 44 (10):  1401-1407.  DOI: 10.12068/j.issn.1005-3026.2023.10.005
    Abstract ( 531 )   HTML ( 16)   PDF (614KB) ( 247 )  
    In order to improve the DOA estimation accuracy of mixed signals and reduce array aperture loss, a high-precision DOA estimation algorithm based on oblique projection operator is proposed. The proposed algorithm estimates the independent and coherent signals of the mixed signals in two stages. Firstly, the covariance matrix of the data received by the array element is processed by the estimating signal parameter via rotational invariance techniques(ESPRIT), and the DOA estimates of the independent signals in the mixed signals are calculated. Then, the algorithm uses the oblique projection operator to remove the independent signal information in the mixed signals to obtain a new covariance matrix. The signal subspace of the newly obtained covariance matrix is used for decoherence processing. Finally, the ESPRIT algorithm is used to calculate the DOA estimates of the coherent signals. Simulation results show that the proposed algorithm has higher accuracy than traditional mixed signal DOA estimation algorithms in the case of low signal-to-noise ratio and small signal incidence interval, and effectively reduces array aperture loss. Under the condition of different sampling snapshot numbers, the proposed algorithm also shows stronger robustness.
    References | Related Articles | Metrics
    A Multi-modal Multi-objective Optimization Algorithm Based on Adaptive Search
    LI Zhan-shan, SONG Zhi-yang, HUA Yun-qiao
    2023, 44 (10):  1408-1415.  DOI: 10.12068/j.issn.1005-3026.2023.10.006
    Abstract ( 543 )   HTML ( 18)   PDF (470KB) ( 234 )  
    The current decomposition-based multi-modal multi-objective optimization algorithms have insufficient population search capability, useless solutions in sub-populations, and a non-universal distance metric. To address these issues, an adaptive search multi-modal multi-objective optimization algorithm MOEA/D-AS is proposed. Firstly, this method increases the number of reference vectors by reducing the size of the average sub-population. Secondly, the sub-populations are reallocated according to the current state of the sub-populations in the iteration. Finally, a clear distance based on local population information is introduced as the basis for modifying the sub-populations. The proposed algorithm is compared with four algorithms on the 2019 CEC multi-modal multi-objective test problems and the large-scale multi-modal multi-objective test problems for experiments. The experimental results show that the proposed algorithm can effectively solve the multi-modal multi-objective optimization problems.
    References | Related Articles | Metrics
    Materials & Metallurgy
    Leaching Kinetics of Magnesiothermic Self-propagating Products in Preparation of Ti6Al4V Powder by Multistage Deep Reduction Method
    YAN Ji-sen, DOU Zhi-he, ZHANG Ting-an
    2023, 44 (10):  1416-1423.  DOI: 10.12068/j.issn.1005-3026.2023.10.007
    Abstract ( 466 )   HTML ( 26)   PDF (1401KB) ( 151 )  
    The leaching process of magnesiothermic self-propagating products in the preparation of Ti6Al4V alloy powder by multi-stage deep reduction method was studied. The effects of raw material particle size, leaching temperature and hydrochloric acid concentration on the leaching process were studied. The results show that temperature and hydrochloric acid concentration have a great influence on the leaching rate of Ti. The apparent activation energy of the Mg leaching reaction was 47.38kJ/mol, and the reaction order was 0.22. The apparent activation energy of Ti leaching reaction was 103.4kJ/mol, and the reaction order was 1.142.Finally,the magnesiothermic self-propagating product with D50 of 59.4μm was selected as the raw material, and the leaching temperature of 30℃ and the hydrochloric acid of 1mol/L were used as the leaching conditions.After 180min of leaching, 92.1% of Mg in the magnesiothermic self-propagating product can be removed, and the loss rate of titanium was 17.5%.
    References | Related Articles | Metrics
    Analysis of Slag Granulation Mechanism and Crushing Efficiency Under Gas Quenching
    LIU Xiao-hong, WEN Zhi, XIAO Yong-li, LOU Guo-feng
    2023, 44 (10):  1424-1430.  DOI: 10.12068/j.issn.1005-3026.2023.10.008
    Abstract ( 522 )   HTML ( 22)   PDF (3792KB) ( 201 )  
    Aiming at the granulation of slag under gas quenching, the numerical simulations of the slag granulation process under gas quenching is carried out on the basis of experiments. The VOF model is used to track the free interface and the Realizable κ-ε model is used to deal with the turbulent flow. The granulation process of gas-quenched slag is simulated and the mechanism of slag fragmentation is analyzed. The dimensionless local momentum ratio is then further established and the influence factors of slag viscosity and gas-slag local momentum ratio are analyzed. The results showed that there are two forms of slag fragmentation under gas quenching. One is the columnar fragmentation caused by R-T instability, mainly caused by local high pressure zones on the windward and leeward sides of the slag flow. The other is surface liquid film breakage dominated by K-H instability, which is mainly caused by the velocity gradient at the gas-slag interface. The increase of slag viscosity will increase the particle size of the slag and decrease the crushing efficiency. The increase of the local momentum ratio enhances the liquid film fluctuations on the surface of the molten slag, and thereby increase the number of stripped particles, reduce the fragmentation length, and enhance the fragmentation efficiency of the molten slag.
    References | Related Articles | Metrics
    Mechanical Engineering
    Dynamic Modeling and Property Analysis of Ball Bearings with Shaft Current Damage
    LI Xiao-peng, QU Xing-chao, LI Bai-tao, SU Jing
    2023, 44 (10):  1431-1439.  DOI: 10.12068/j.issn.1005-3026.2023.10.009
    Abstract ( 569 )   HTML ( 21)   PDF (2359KB) ( 150 )  
    Aimed at the wind turbine ball bearings, a shaft current damage model in the form of pits is proposed to investigate the influence of the shaft current pit damage on the bearings’ dynamic characteristics. A damage model with length, width and depth is established based on the half-sine and piecewise excitation functions to describe the different damage degrees of the shaft current. The dynamic model of the bearings with shaft current damage is built to study the influence of the pit size, the pit shape and the external load on the vibration properties of the bearings. The effect of the shaft current damage with different damage degrees on the load distribution of the bearings is investigated by the contact analysis. The correctness of the proposed model is verified by comparing the results from the numerical simulation and the experiment. The results show that the shock excitation and the periodic phenomenon caused by the shaft current pit damage can be accurately described by the proposed model. With the shaft current damage increasing, the vibration response and the contact force of the bearings increase, and the stability becomes worse.
    References | Related Articles | Metrics
    Probability Analysis of Zero Clearance Position of Wind Turbine Spindle Bearings
    HUANG Xian-zhen, ZHANG Peng, LI Hong-lei, LYU Zhong
    2023, 44 (10):  1440-1447.  DOI: 10.12068/j.issn.1005-3026.2023.10.010
    Abstract ( 442 )   HTML ( 15)   PDF (1495KB) ( 109 )  
    The preload greatly affects the operating state and service life of bearings. To ensure the reasonable assembly of the spindle bearings, the position of zero clearance for the bearings needs to be determined. Firstly, the variation of clearance is calculated by using the thick-walled cylinder theory to determine the zero clearance position of the bearings. Then, the finite element displacement model of bearing zero clearance is established, and a more accurate calculation method of zero clearance position is proposed. Considering the influence of random factors, the Kriging surrogate model is proposed to analyze the probability distribution characteristics of the inner and outer ring positions under the zero clearance of wind turbine spindle bearings. Finally, numerical examples show that the maximum error predicted by the Kriging model is within 0.1%, indicating that the proposed method has higher accuracy and applicability.
    References | Related Articles | Metrics
    Motion Mechanism and Test Study of Vibration-Driven Robot with External Fins
    GUO Yu-liang, GUO Rui, TANG Chen-wei, YAO Hong-liang
    2023, 44 (10):  1448-1454.  DOI: 10.12068/j.issn.1005-3026.2023.10.011
    Abstract ( 523 )   HTML ( 18)   PDF (2384KB) ( 105 )  
    To satisfy the demand of vertical pipe inspection, a vibration-driven robot with external fins is designed and the motion mechanism is studied. Based on the structure of the robot, the non-smooth multibody dynamics model of the robot system is developed, and the two non-smooth nonlinearities, which are friction and impact, are considered. Numerical simulation analysis of the frequency and time domain response of the robot are carried out and the effects of different load quality parameters on the robot are studied. The robot prototype is developed, including microcontrollers, wireless control modules, etc. The robot is tested for vertical pipe crawling, and additional mass is added to the robot to verify the load capacity of the robot during the test. The test results verify the correctness of the numerical simulation results and the maximum upward crawling speed of the robot is 10.8mm/s.
    References | Related Articles | Metrics
    Strain Reconstruction of TBM Cutterhead at Key Positions Based on BP Neural Network
    HUO Jun-zhou, GE Li-han, ZHANG Zhan-ge, LI Gao-rui
    2023, 44 (10):  1455-1463.  DOI: 10.12068/j.issn.1005-3026.2023.10.012
    Abstract ( 371 )   HTML ( 12)   PDF (2200KB) ( 118 )  
    To solve the problem of difficult and inaccurate strain monitoring of the key main load-bearing structures of the full face hard rock tunnel boring machine(TBM)under harsh working conditions, a strain reconstruction method for key positions of the TBM cutterhead based on BP neural network and finite element analysis was proposed. Firstly, the key vulnerable positions of the TBM cutterhead were determined through static and dynamic finite element analysis, and the typical vulnerable feature substructures of the cutterhead were extracted. Secondly, the static finite element analysis of standard parts and feature substructures under multiple loads was carried out based on the design of experiments(DOE), and the load-strain database was constructed.Finally, a strain reconstruction model for the standard sample and cutterhead feature substructure was established using BP neural network, and experimental verification of the standard samples were conducted.The results showed that the average error of the reconstructed strains is 10%, which verifies the feasibility of the method and provides a feasible method for strain reconstruction of complex TBM cutterhead.
    References | Related Articles | Metrics
    Classification and Identification of Excavators’ Working Stages Based on Deep Learning
    LIU Wei-wei, DENG Jian-yang, ZHANG Jing-wen, NIU Dong-dong
    2023, 44 (10):  1464-1474.  DOI: 10.12068/j.issn.1005-3026.2023.10.013
    Abstract ( 464 )   HTML ( 22)   PDF (1788KB) ( 216 )  
    In order to realize the automatic identification of each working stage of excavators’ operation cycle, an intelligent identification method is adopted which takes the pilot pressure of actuators, and the pressure and power of the main pump as the identification objects. The working stages were divided by the pilot pressure change of each actuator, and then verified by the pressure and power change of the main pump. Defining the waveform that begins with each working stage as the staged symbol, the data features were extracted and the optimal time window width was determined in the form of time-window slips. Deep learning was used to identify each segment marker. Comparing the identification effects of ResNet and LSTM neural networks, which are widely used in the field of classification identification in deep learning, it was found that LSTM has better identification effects, and the identification accuracy of the test set can reach 99.75%. LSTM is used to identify the test data, and the identification accuracy is only 82.54%, indicating that there exists misidentification. Based on the logical sequence of excavators’ working stages and the power threshold of the main pump, the identification accuracy can be increased to 99.72%. The results show that the proposed method has high identification accuracy and good real-time performance, and can effectively identify each working stage of the operation cycle.
    References | Related Articles | Metrics
    Improved SSD Rapid Separation Model of Coal Gangue Based on Deep Learning and Light-Weighting
    LI Juan-li, WEI Dai-liang, LI Bo, WEN Xiao
    2023, 44 (10):  1474-1480.  DOI: 10.12068/j.issn.1005-3026.2023.10.014
    Abstract ( 428 )   HTML ( 16)   PDF (1626KB) ( 122 )  
    A new model DSR-SSD for coal gangue fast identification is proposed based on the SSD model to address the issues of large parameter quantities and low operating speed in the SSD model. The application of deep separable convolutions in the backbone feature extraction network reduces the computational complexity, and integrating the RFB module into the SSD model improves the model’s feature extraction ability. After verification, the recognition rate of the DSR-SSD model is 113.99 frames/s, and the accuracy rate is 95.17%. Comparing DSR-SSD with SSD, Faster-RCNN, and YOLOv3 models, it is found that the DSR-SSD model improves the accuracy by 2.29% and the recognition rate by 60.89% compared to the SSD model, and the accuracy of the DSR-SSD model is 2.86% higher than the Faster-RCNN and 2.71% higher than the YOLOv3, with recognition rates 14.90 and 3.65 times higher than the Faster-RCNN and YOLOv3.
    References | Related Articles | Metrics
    Resources & Civil Engineering
    Hybrid Recognition Model of Microseismic Signals for Mining Based on Mel Spectrum and LSTM-DCNN
    ZHAO Yong, JIAO Shi-hui, ZHAO Qian-bai
    2023, 44 (10):  1481-1489.  DOI: 10.12068/j.issn.1005-3026.2023.10.015
    Abstract ( 492 )   HTML ( 14)   PDF (1547KB) ( 145 )  
    Microseismic monitoring can ensure safe production in mines, and the accuracy of microseismic signal recognition directly affects the analysis of microseismic events. The microseismic monitoring data of Xiadian Gold Mine were used as samples to establish a mining microseismic signal recognition model based on Mel spectrum and a combination of long short-term memory(LSTM)and deep convolutional neural networks(DCNN). Firstly, the monitoring signal was preprocessed, and the Mel time spectrum was used to reduce the weight of the interference frequency band and sample size. Then, LSTM and DCNN models were employed to extract the temporal and spatial features of the signal, respectively. Through comparative analysis of multiple models, the results showed that the proposed Mel-LSTM-DCNN hybrid model has the highest accuracy in identifying microseismic signals. The model proposed provides reference for accurately identifying microseismic signals in mines.
    References | Related Articles | Metrics
    Numerical Simulation of Soil Deformation During Jacking of Circular Steel Pipes with Flange Plates
    ZHAO Wen, WANG Zhi-guo, WANG Zhao-peng, WANG Xin
    2023, 44 (10):  1490-1498.  DOI: 10.12068/j.issn.1005-3026.2023.10.016
    Abstract ( 512 )   HTML ( 14)   PDF (1457KB) ( 103 )  
    In the traditional pipe curtain construction, two steel plates, known as flange plates, are welded on both sides of the original circular steel pipe, creating a circular steel pipe with flange plates. Adjacent steel pipe with flange plates are welded together to form a new pipe curtain structure. The presentence of the flange plate cause the soil deformation law during jacking process to differ from that of circular steel pipes. In this study, Midas GTS finite element analysis software was used to simulate the jacking process of circular steel pipes with flange plates of different pipe diameters and flange plate positions. The results show that the soil settlement at the pipe top, the soil uplift at the bottom of the pipe, and the vertical displacement of the soil at the flange plate increase continuously with an increase in the spacing between the upper and lower flange plates or an increase in pipe diameter. When the top excavation surface is one section length away from the monitoring section, soil uplift occurs at the monitoring section position. When the top reaches the monitoring section position, the soil deformation becomes settlement. When the top position exceeds one section length of the monitoring section, the soil deformation tends to stablize. Meanwhile, the functional relationship between soil deformation and spacing between upper and lower flange plates, as well as pipe diameter, is obtained by fitting the curve.
    References | Related Articles | Metrics
    Development Model of Concrete Corrosion Damage in Multi-salt Coupling Environment
    HE Sheng, WANG Xiao, YU Peng, LI Yu-tao
    2023, 44 (10):  1499-1506.  DOI: 10.12068/j.issn.1005-3026.2023.10.017
    Abstract ( 493 )   HTML ( 14)   PDF (837KB) ( 142 )  
    The solution used in this study contained mainly SO2-4, Cl-, and Mg2+ ions, which closely resembled the chemical corrosion environment in the factories. Concrete degradation was investigated by conducting on-site exposure tests and indoor simulation tests, studying the influences of different soaking method, water binder mass ratio, solution mass fraction, and dry-wet cycle on concrete degradation. Ultrasonic wave detection was used to analyze the degree of concrete degradation. The results showed that the thickness of concrete damaged layer increased due to the reaction, however, the growth rate decreased gradually under the multi-salt coupling corrosion. In the early stage of corrosion, Cl- and SO2-4 entered the concrete and reacted with each other, causing a greater concentration difference between the inside and outside of concrete, which accelerated the invasion of corrosion ion and the corrosion rate. In the middle and late stage of corrosion, Mg2+ and SO2-4 participated in the corrosion, leading to increased defects. As the corrosion progressed, the concentration difference decreased, and the corrosion rate slowed down. Based on the experimental results, a damage development model was proposed, and it was concluded that the indoor simulation test corrosion of 1.37d was equivalent to the on-site semi-immersion test corrosion of 1d, with relatively accurate results.
    References | Related Articles | Metrics
    Management Science
    Method for Hotel Partner Selection Considering Demand Information and Electronic Word-of-Mouth
    YOU Tian-hui, ZHANG Xi-ting, CAO Bing-bing, YUAN Yuan
    2023, 44 (10):  1506-1513.  DOI: 10.12068/j.issn.1005-3026.2023.10.018
    Abstract ( 411 )   HTML ( 14)   PDF (477KB) ( 106 )  
    Considering that travel agencies will pay attention to the demand information and electronic word-of-mouth(e-WOM)development trend of each alternative hotel when choosing partners, a method for hotel partner selection is proposed based on both information. First, the directed and weighted graph is constructed, in which the weight of the node is determined based on the demand information. Further, a time weight correction coefficient that considers the horizontal and vertical development trends of the alternatives’ e-WOM information is proposed to calculate the weight of each stage, and the TOWGA operator is used to calculate the attribute performance of each alternative hotel. By comparing the e-WOM attribute performance, the directed edge and its weight are determined. Then, the algorithm for calculating the ranking value of the alternative hotels is given based on PageRank. Finally, an example is used to verify the feasibility and effectiveness of the proposed method.
    References | Related Articles | Metrics
    Study on Visual Search Performance of Cockpit Interfaces Under Time Pressures
    ZHOU Yao, CHEN Deng-kai, TAN Xiao-xue, ZHAO Min
    2023, 44 (10):  1514-1520.  DOI: 10.12068/j.issn.1005-3026.2023.10.019
    Abstract ( 350 )   HTML ( 20)   PDF (4839KB) ( 163 )  
    To investigate the search characteristics of aircraft cockpit interfaces under different time pressures to improve task performance, the visual search trajectories and task performance data of the subjects were obtained by simulating the task search process under aircraft flight conditions, using such methods as eye tracking, performance evaluation and scale measurement, and then the multi-channel data obtained were analyzed. The results showed that there are significant differences in the visual search characteristics of the subjects under different time pressures(none, low, high) and different interface complexity(simple, complex); both high time pressures and complex interface displays placed a high cognitive load on the subjects; and there is an inverted U-shaped relationship between time pressures in the simple interface displays. It provides a scientific basis for the design of complex human-machine interfaces in aircraft cockpits.
    References | Related Articles | Metrics