2022 volumne 43卷 Issue 06
YANG Qi1, LIU Mugeng2, MA Yun3
Abstract: UI manuscript recognition is one of the important applications of image object detection in the area of software engineering. Due to the significant difference between UI manuscript images and natural images where UI manuscript images usually need to be drawn manually, it is difficult to build UI manuscript dataset for deep learning because of the dependency on tremendous manual efforts. To address the issue, in this study an approach called UIsketcher was proposed to efficiently generate UI manuscript dataset based on optimizing the current workflow. In UIsketcher, users should just draw some basic elements without labeling, and then the dataset could be automatically generated for training deep learning model. According to the experiment with UIsketcher, only 25% drawing workload of the traditional methods could get the similar training results. If the workload was 75%, the final accuracy was even better than that of traditional methods.
WANG Peichong1,2, YIN Xinjie1,2, LI Lirong3
Abstract: To overcome the weakness of slow convergence, prematureness and low accuracy of seagull optimization algorithm (SOA) in solving high-dimensional problems,an improved SOA with learning (ISAOL)was proposed. A migration operator based on the difference between the Xi and the Xmwas designed, this could make Xi search wider solution spaces in early stage, and a nonlinear adaptive parameter A was introduced to ensure the algorithm suitable for the search of solution space of complex problems, which could prevent the algorithm from falling into local optimum too early. In the later stage, some elite individuals executed opposition based learning(OBL) to intensify the exploration of the space around the global optimal individual to improve the accuracy of solution. Ten unconstrained test functions in CEC2017 were selected to test the performance of the ISOA and compared with HPSO-TS, V-DVGA, DADE, CMA-ES and others. Testing results of this experiments showed that the ISOAL had higher accuracy and stability than other algorithms. Finally, experiments was carried out by using the tension spring problem. The results showed that the total cost of the spring , the coil diameter and the average diameter of the spring obtained by the ISOAL were reduced by 3.5% ,5.7% and 3.5% than SOA, respectively. ISOAL had the attributes of fast convergence, high accuracy and robustness, fitting to solve higher dimensional function optimization problem and engineering optimization problems with constraints.
HUANG Dongqiang, HUANG Jianqiang, JIA Jinfang, WU Li, LIU Lingbin, WANG Xiaoying
Abstract: In order to improve the computational efficiency of the GRAPES(global/regional assimilation and prediction system) numerical weather prediction model, and to improve the performance of the dynamic framework, In order to solve the problem that the GCR algorithm was time-consuming in GRAPES mode, a CPU+GPU heterogeneous parallel preprocessing GCR algorithm was implemented. Firstly, incomplete LU decomposition was used to preprocess the coefficient matrix to reduce the number of iterations. On this basis, fine-grained parallelism of OpenMP and coarse-grained parallelism of MPI were implemented. OpenMP parallelism was mainly used to improve the performance of the program by using compiler guidance to the loop body without data dependence in the way of loop unrolling. MPI parallelism was used to divide the data into various processes and improve the scalability of parallel programs by non-blocking communication and optimizing the amount of communication data. MPI was responsible for process communication and iterative control between nodes, while CUDA was responsible for processing computation-intensive tasks. The time-consuming matrix calculation part of GCR was transferred to GPU for processing, and memory optimization and data transmission optimization were adopted to reduce the data transmission overhead between CPU and GPU. The experimental results showed that the parallel acceleration ratio of OpenMP was 2.24 times that of the serial program, the parallel acceleration ratio of MPI was 3.32 times that of the serial program, and the parallel acceleration ratio of MPI+CUDA was 4.69 times that of the serial program. The performance optimization of the generalized conjugate redundancy algorithm on the heterogeneous platform was realized, and the computational efficiency of the program was improved.
HUANG Hua1,2,JIANG Jun3,YANG Yongkang2, HE Defeng1,CAO Bin4
Abstract: In this study, the service function chain (SFC) orchestration problem in data centers was investigated. The relationship between request arrival rate, computing resources and processing delay were analized. A flexible resource allocation optimization model aiming at maximize deployment benefits was proposed. Moreover, a heuristic method is developed to solve the problem in two stages. Firstly, the resource usage for SFC was optimized based on flexible resource allocation considering service delay. Secondly, for the virtual network function deployment and link mapping, the top-down and bottom-up search strategies based on worst-fit strategy were alternately adopted to improve the deployment efficiency and reduce the SFC delay. Finally, simulation with 4 and 6 pods in data center was designed to verify the performance of our heuristic algorithm. Experimental results demonstrated that, compared with existing methods, the development profit, success rate and resource utilization were improved in our algorithm.
FENG Haonan, HE Zhiyong, MA Liangli
Abstract: In order to solve the information overload problem on social media platforms and help users quickly capture the required information, in this study the problem of hashtag recommendation based on multimodal content was investigated. To address the heterogeneous differences between different modalities, a co-attention mechanism was used to model and fuse features of cross-modal content, and use Seq2Seq framework was used to generate new hashtag sequences to address the deficiency that multi-label classification methods could only recommend hashtags in the hashtag space of the dataset. An aggregation strategy was used to aggregate the recommendation results of classification methods into the generated hashtag sequences to obtain a unified recommendation model for both methods. The experimental results on a large-scale dataset showed that, firstly, the multimodal approach was more advantageous than the unimodal approach, and the unified recommendation model proposed in this paper had 9.44 percentage points improvement in F1 value over the comparison model using unimodal approach, and 3.41 percentage points improvement over the comparison model using the classification method. Finally, the unified recommendation model UNIFIED-CO-ATT is 1.25 percentage points higher than GEN-CO-ATT in F1 values. The model proposed in this study could combine the advantages of classification and generation methods and could make the recommended hashtags have the advantages of accuracy and novelty at the same time.
ZHOU Aohui1, WENG Zhiyuan2, ZHOU Siyuan1, HUANG Qiao1, WANG Ye1, ZHANG Hua1
Abstract: Among the large number of available services on the internet, how to efficiently match the right service for a specific business target is a major challenge in current research. To address this problem, a method based on topic fitering and semantic matching was proposed that could be used for massive service discovery. The method first used Word2Vec to compare the similarity between the topic description text and the business target description text to obtain the business target topic, and then used TextRank to extract the service key sentences from the service description text. The service key sentences were filtered by the extracted business target topics to narrow the comparison range. Then, the word vector was extracted from the corresponding business goal and service description text, and the BiLSTM model with attention mechanism was used to calculate the similarity between them and return the list of the TOP-N services that were most similar to the business target description text for business developers for selection. And the data crawled from Programmable Web was annotated to build the business target-service sentence dataset required for the experiments, and evaluate the effectiveness of the methods in this study. Finally, the comparison results with models such as TextCNN , BiLSTM, and Word2VecSD showed that MAP of this method could be increase by 1.41 percentage points, 4.61 percentage points, and 4.95 percentage points. The finding of this study lay solid ground for further improvement in future work.
WANG Junfen1,2, LIU Peiyue3, DONG Jianbin1, 2, ZHU Zhanlong1
Abstract: The Fuzzy C means algorithm cannot effectively segment object pixels from a non-destructive testing (NDT) image because of the great difference between the background and the object region. Therefore, a fast fuzzy C means algorithm for segmentation of NDT image was proposed in this study. In the process of clustering, local spatial information and gray information were introduced to improve the robustness of the algorithm. The condition value was used to represent the sample size to balance clusters of different sizes and solve the cluster-size sensitivity problem. Based on the new constraints, a new form of membership degree and cluster center representation could be obtained. The execution time of image segmentation was no longer determined by the size of the image, but by the gray level of the image. The computational complexity of the algorithm was greatly reduced. NDT images and synthetic images with a large difference in cluster size were used for testing. The segmentation accuracy (SA), F-value, G-mean and the execution time of image segmentation were used to evaluate the performance of algorithms. The experimental results showed that when the original test image was polluted by Gaussian noise, salt and pepper noise, Rayleigh noise and multiplicative noise, the proposed algorithm had better robustness, SA is 97.93%, F-value is 88.50%, G-mean is 93.83%, the execution time of image segmentation was less, about 14.06 ms. The simulation tests could verify the effectiveness of the proposed algorithm.
DU Zhiye1, ZHAO Pengfei1, WU Chuan2, LONG Zhenhua3, MENG Shengchun1, HAO Zhaoyang1
Abstract: Aiming at the problem of large measurement errors in the current stress-strain monitoring methods of transmission tower, this study was carried out to improve the accuracy of strain measurement of transmission pole and tower angle steel. Based on fiber bragg grating (FBG) sensor, an accurate online monitoring method for angle steel strain of transmission towers was proposed considering the position and mode of sensor installation on the surface of angle steel. Firstly, the transmission tower angle steel surface strain distribution characteristics of different positions were analized based on the finite element simulation, and then according to the transmission tower structure in L shape characteristics of angle steel, a removable hand ring type L-type joint angle sensor was designed on the surface of the fixed clamp. The comparison and analysis of different sensor installation methods and their influence on the angle steel strain measurement accuracy were examined. Finally, this method was used to design an accurate online monitoring system of transmission pole tower angle steel strain based on fiber bragg grating sensor. The strain data of angle steel in weak position of pole tower of a 220 kV transmission line was measured and analyzed. The results show that the strain measurement error of the measured data and the finite element simulation data was less than 1.5×10-5; and the relative error was less than 6.28%, which could effectively improve the measurement accuracy of the Angle steel strain of transmission tower. The finding of this study showed extra light on the accurate online monitoring of transmission tower structure.
ZHAO Guifeng,WEI Danyang,ZHANG Meng
Abstract: Most of the existing simulation studies on the wind-induced vibration of ice-coated conductors simplify the conductor to an ideal smooth circular section and accept its aerodynamic characteristics, but the section characteristics of the actual conductor twist are not considered enough. In this study, Fluent software is used to study the aerodynamic characteristics of ice-coated smooth circular and non smooth circular twisted section conductors, and to analyze the characteristics and differences of their aerodynamic coefficients at different wind speeds, icing thickness, icing shape and conductor diameter. The results showed that the aerodynamic coefficients of ice-coated conductors with two sections were quite different, among which the maximum relative deviation of drag coefficient and torsion coefficient could reach 100%, and the maximum relative deviation of lift coefficient could reach -175%.The use of non smooth circular twisted section conductor could increase the number of vortex shedding in the cycle and made its fluctuation more violent. When the icing was crescent shaped, the smooth circular section conductor could underestimate the downwind displacement and overestimate the crosswind displacement of the conductor. When the icing was D-shaped, the fluctuation of smooth circular section conductor was smoother and more regular than that of non smooth circular twisted section, which could overestimate the downwind displacement and underestimate the crosswind displacement. In view of the large differences in aerodynamic force, vortex shedding diagram and displacement response between the two types of coated ice conductors, the influence of conductor cross-section stranded characteristics on aerodynamic force should be considered in the fine wind resistance design of coated ice transmission line system.
LIU Huadong, XU Haoxuan, LI Hewei, WANG Dingbiao
Abstract: In order to realize the rapid and simple detection of Fe3+ and Hg2+ in aqueous solution, sulfur and nitrogen co-doped carbon quantum dots fluorescent probes were synthesized by one-step hydrothermal method using citric acid and thiourea as precursors. The morphology and fluorescence properties of the N—S—CDs were analyzed by transmission electron microscope, Fourier transform infrared spectroscopy, X-ray photoelectron spectroscopy, fluorescence spectrophotometer, and ultraviolet-visible absorption spectroscopy. The results showed that the average particle size of N—S—CDs was 8 nm, which was uniformly distributed. The surface was rich in various functional groups and has good water solubility. The strong blue fluorescence was emitted from N—S—CDs under the irradiation of ultraviolet lamp, and the fluorescence quantum yield was as high as 36.8%. With the increase of excitation wavelength, the fluorescence intensity of N—S—CDs emission peak increased at first and then decreased, and the fluorescence intensity reached the highest at 440 nm, and the emission peak was obviously red-shifted with the excitation wavelength. N—S—CDs had high selectivity for Fe3+ and Hg2+. There was a good linear relationship between the fluorescence intensity and the two ions when the concentration of Fe3+ was in the range of 40—130 μmol/L and the concentration of Hg2+ was in the range of 40—80 μmol/L.The linear regression equations are F0/F=0.021 32cFe3++ 0.349 45 and F0/F=0.186 11cHg2+-6.271 14, and the lowest detection limits were 1.4 μmol/L and 0.16 μmol/L, respectively. The application of N—S—CDs to the real sample analysis of tap water and lake water could effectively detect Fe3+ and Hg2+. The quantitative recovery of Fe3+ was between 97.16% and 103.62%, and that of Hg2+ was between 93.58% and 101.22%.
YUN Zimeng, LIU dandan, HUANG Jiajia
Abstract: Owing to the large radius of potassium ions, the SPAN cathode suffers large volume change during the cycling process, which causes structure collapse and serious capacity fade. Herein, the effects of sodium carboxymethyl cellulose (CMC) and polyvinylidene fluoride (PVDF) on the electrochemical performance of SPAN cathode in K-S battery were investigated, and the reason of CMC binder improving the electrochemical performance of SPAN cathode was further explored by kinetic test and SEM test. The rate performance test results showed that the SPAN electrode prepared by CMC binder displayed capacity of 1 256, 1 161, 1 058, 946, 716, 538 mAh/g at 0.1C, 0.2C, 0.5C, 1C, 2C, 3C. And the capacity remained 1 253 mAh/g when the current density returns to 0.1C, which was far better than the rate performance of SPAN electrode prepared by PVDF binder. Besides, the cycle performance test results showed that the SPAN electrode prepared by CMC binder could deliver a high reversible capacity of 822 mAh/g after 100 cycles at 0.5C, corresponding to a high capacity retention of 75.6%. The mechanism test results showed that benefit from the abundant oxygen-containing active functional groups in the CMC structure which could not only strengthen chemical interactions with active materials, alleviate the volume change of SPAN cathode during the charge/discharge process, but also accelerate the transport of potassium ion. The SPAN electrode prepared by CMC binder demonstrated a high rate performance and cycle performance compared to the case of PVDF binder.The finding of this study provided an effective strategy to restrain the volume expansion of potassium-sulfur batteries, to maintain the stability of the conductive network inside the electrode, thereby to improve the rate performance and cycle stability.
PENG Zhaoxu, WANG Ju, LOU Tianyu, JIANG Kun, NIU Ningqi
Abstract: In order to explore the impact of phosphorus shock load on the characteristics of activated sludge system, the anaerobic aerobic SBR was used for experiments. By changing the influent phosphorus content, the pollutant removal characteristics and sludge sedimentation performance of activated sludge system were investigated in the conditions of influent carbon-phosphorus ratios of 330/8, 330/12, 330/16, 330/20. The results showed that the decrease of carbon-phosphorus ratios could enhance the activity of phosphorus accumulating bacteria, could improve the sludge settleability, and could significantly improve the system′s nitrogen and phosphorus removal performance. When the influent carbon-phosphorus ratios changed from 330/8 to 330/20, the specific phosphorus uptake in the aerobic stage increased from 9.502 mg/g to 17.764 mg/g, increasing by 86.95%. With the impact of increasing phosphorus concentration, the phosphorus accumulating bacteria would absorb more organic matter by anaerobic phosphorus releasing, and the experimental effluent quality will be improved. During the anaerobic period, the pH decreasing rate was significantly correlated with the phosphorus release rate (R2=0.667), and the pH curve reflected the characteristics of anaerobic respiration in the system. ORP decreased continuously in the anaerobic stage, and two platform appeared in the aerobic stage. The change process of PO43--P concentration could be indicated by online monitoring of ORP changes, and the time point of the end of anaerobic phosphorus release could be determined. When the influent COD remained unchanged, increasing the influent phosphorus concentration could enhance the microbial activity, improve the sludge sedimentation performance and the system′s nitrogen and phosphorus removal performance, which would have a favorable impact on the activated sludge system.
YANG Ping, WANG Yanzi, DIAO Pengcheng
Abstract: In the prediction and control of frost heave and thaw settlement, the effect of external load, especially large load, on inhibiting frost heave and the law of frost heave deformation are the key to predict and control frost heave deformation. Through the frost heave test of remolded silty clay in the condition of water replenishment, the influence and evolution law of different load levels on the frost heave characteristics of remolded silty clay were studied. The results showed that the cooling law of samples with different loads was similar, which was mainly divided into four stages: constant temperature, rapid temperature drop, slow temperature drop and temperature stability. The overall performance was that the greater the load, the faster the temperature drop. The development curves of freezing front with time with different loads showed a logarithmic growth trend. With the increase of load, the development speed of freezing front in the early stage of freezing was accelerated, and the development speed of front in the middle and late stages was slowed down. The height of front position decreased with the increase of load. With the increase of load, the time when the initial frost heaving amount appeared delayed, the time when the frost heaving amount reached stable development shortened, the development rate of frost heaving amount decreased, and the frost heaving amount decreased. The final frost heaving amount of soil sample without load was 6.4 times that under 200 kPa. The frost heaving ratio decreased exponentially with the increase of the overlying load, and the final frost heaving ratio of soil samples with no load was 5.6 times that under 200 kPa.
LI Aimin 1, WANG Hailong2, XU Youcheng 2
Abstract: In remote sensing monitoring of dissolved organic carbon (DOC) content in urban lakes, the traditional regression model is difficult to describe the nonlinear relationship and can not meet the accuracy requirements. In this study, Bayesian optimization algorithm was introduced into the parameter optimization of random forest model, and a DOC concentration inversion method of urban lakes based on Bayesian optimization random forest model (BO-RF) was proposed. Taking the water area of Tiande Lake in Zhengzhou city as an example, the remote sensing inversion method of DOC concentration in urban lakes was studied based on high spatial-temporal resolution Planet satellite image data and measured DOC water quality data. Through PEARSON correlation analysis, the results showed that the best band combination of Planet satellite image band for retrieving DOC concentration was B2/B4. The determination coefficient R2 of the band ratio model obtained by the traditional regression method was 0.466, and the root mean square error RMSE was 0.515 mg/L, which could not meet the accuracy requirements. The modeling accuracy was improved by using support vector machine and BP neural network, the fitting R2 was 0.772 and 0.806, respectively, and the root mean square error RMSE was 0.328 mg/L and 0.302 mg/L, respectively. Bayesian optimization algorithm was introduced to optimize the random forest model to obtain the BO-RF model, and its fitting degree R2 was 0.865 and root mean square error RMSE was 0.253 mg/L. The BO-RF fit of the optimized model was good, and the accuracy of the model was significantly improved. The Bayesian optimized random forest BO-RF algorithm was more suitable for retrieving DOC concentration in Tiande lake, which could provide a reference for remote sensing monitoring of urban lake water quality.
YUAN Chengfang1, CHEN Yang1, WANG Shibo2, WEI Yiran1
Abstract: The recycled micropowder formed by crushing and ball milling waste clay bricks has potential activity. It can partially replace cement to prepare concrete through activation excitation, so as to realize the reuse of waste resources. This study explored the excitation effect of mechanical excitation, chemical excitation and high temperature excitation on the activity of recycled brick micropowder, and SEM and XRD technology analysis were used to reveal the material microstructure and mechanism of action. The results showed that after using mechanical excitation, chemical excitation, and high temperature excitation to activate the regenerated micropowder, the activity index of the material was improved to varying degrees, and there was a good correlation between different excitation methods and the activity index. Among them, the 45 min mechanical excitation of ball milling and the high temperature excitation of 800 ℃ had better effects, with the highest activity index reaching 71%, and the highest activity index in the chemical excitation mode was 65%. Considering the excitation effect, energy consumption, economy and practical operation feasibility, it is recommended to use ball milling for 45 min as the best activation method for recycled brick micropowder.
ZHAO Yi1, DUAN Songfu1, NIU Zhonghao1,2
Abstract: In order to achieve the seismic performance and overall aesthetic feature of special-shaped concrete-filled steel tubular column joints, a new side plate connecting T-shaped steel tubular special-shaped column-steel beam node was proposed. The experimental study of seven T-shaped concrete-filled steel tubular column-steel beam joints with side plates under low cyclic loading was carried out. The influencing factors such as the presence or absence of side plates, the length of side plates, the presence or absence of restrained tie rods, and the axial compression ratio were examined. The failure process of the node was observed, the hysteresis curve and skeleton curve of specimens were recorded. The experimental results showed that the failure modes of specimens were divided into beam hinge failure with side plates connection specimens and column hinge failure without side plates connection specimens. The hysteretic curve of the new type of concrete-filled steel tubular column and steel beam joints connected by side plates were full and the seismic performance was well. Based on experimental results, the limit load and its corresponding displacement for dimensionless processing was used to obtain the three fold skeleton curve model, all the specimens′ stiffness degradation law and hysteresis curve paths were analyzed, and the restoring force model of the new T-shaped concrete-filled steel tubular column-steel beam joint was established, contrast experimental results verified the correctness of the proposed restoring force model.The finding of this study a foundation for the in-depth study of the seismic performance of such joints.
Copyright © 2023 Editorial Board of Journal of Zhengzhou University (Engineering Science)