Loading...

Table of Content

    25 February 2024, Volume 52 Issue 2
    2024, 52(2):  0. 
    Asbtract ( 18 )   PDF (18199KB) ( 115 )  
    Related Articles | Metrics
    Computer Science & Technology
    LU Yiqin, HUANG Chenghai, CHEN Jiarui, et al
    2024, 52(2):  1-12.  doi:10.12141/j.issn.1000-565X.230032
    Asbtract ( 730 )   HTML ( 2)   PDF (4196KB) ( 30 )  
    Figures and Tables | References | Related Articles | Metrics

    With the progress of network technology, applications such as vehicle networks, industrial Internet of Things and 5G ultra-reliable low-delay communication (uRLLC) all require TSN to ensure ultra-low delay deterministic data transmission. TSN traffic scheduling requires a fast and accurate scheduling algorithm. The existing accurate solution methods are of high complexity and cannot meet the real-time requirements in large-scale joint scheduling. This paper designed a routing optimization genetic algorithm (Routing-GA) with better performance. Combining routing and traffic scheduling constraints, it can improve the efficiency of scheduling algorithm by optimizing routing and provide services for link load balancing scheduling. This strategy increases the space and flexibility of scheduling, and has the characteristics of fast near-optimal solution of meta-heuristic algorithm. It can deal with large-scale TSN routing constraint joint scheduling problem simply and effectively. Routing-GA takes the minimum end-to-end delay of time-sensitive flow as the optimization objective, considers Routing and TSN constraints jointly, and provides a genetic algorithm coding method with low complexity, high efficiency and high scalability according to the characteristics of TSN transmission problems. In addition, in order to improve the performance of the scheduling algorithm, a crossover mutation mechanism was proposed to optimize the route length and link load balancing. The experimental results show that the realized Routing-GA can effectively reduce the end-to-end delay and significantly improve the solution quality. The evolution rate can reach 24.42%, and the average iteration time of traditional genetic algorithm (GA) is only 12%. It can effectively improve the performance of the algorithm and meet the constraint requirements of TSN scheduling.

    WU Haotian, REN Changqing, LU Lu, et al
    2024, 52(2):  13-22.  doi:10.12141/j.issn.1000-565X.230066
    Asbtract ( 100 )   HTML ( 4)   PDF (1989KB) ( 54 )  
    Figures and Tables | References | Related Articles | Metrics

    Given the faster speed of low-precision floating point operations, more and more high-performance applications are using hybrid precision solutions to accelerate.The large AI (artificial intelligence) models that use this scheme to accelerate has also received wide attention. Recently, the HPL-AI (High Performance LINPACK for Accelerator Introspection) benchmark has been proposed to evaluate the mixed-precision computing performance of high-performance systems. For this benchmark test, this study designed and optimized the implementation of single-node HPL-AI benchmark test on Kunpeng and Ascend heterogeneous platforms. In order to balance the load of the AI processor, the tasks were evenly distributed to the AI processors through the cyclic task allocation strategy. The task allocation strategy with interval value was used to improve the continuity of data transmission to reduce the data transmission time between CPU and AI processor. Without affecting the calculation accuracy, the computation on the CPU side was reduced by the strategy of canceling the data scaling. The final experimental results show that the HPL-AI benchmark has the fastest mixed-precision floating-point arithmetic speed when the interval value is 8; at the same time, unscaling the data does not affect the accuracy of the HPL-AI benchmark results. Compared with the non-optimized HPL-AI benchmark implementation on the heterogeneous platform of Kunpeng and Ascend, the optimization strategy proposed in this paper improves the mixed-precision floating-point arithmetic speed by about 29%, which lays a solid foundation for the further optimization of single-node HPL-AI benchmark and the deployment of multi-node HPL-AI benchmark.

    FU Pengbin, XU Yu, YANG Huirong
    2024, 52(2):  23-31.  doi:10.12141/j.issn.1000-565X.230034
    Asbtract ( 100 )   HTML ( 3)   PDF (1012KB) ( 34 )  
    Figures and Tables | References | Related Articles | Metrics

    With complex two-dimensional structure, offline handwritten mathematical expressions is difficult to recognize due to the variable scale of their symbols and the various transformation of their writing styles. This paper proposed a mutual learning model based on multi-scale feature fusion. Firstly, to enhance the model for extracting fine-grained information from expressions and comprehending semantic information of global two-dimensional structures, multi-scale feature fusion was introduced in the encoding stage. Secondly, paired handwritten and printed mathematical expressions were introduced for training the mutual learning model, which includes decoder loss and context matching loss to learn LaTeX grammar as well as semantic invariance between handwritten and printed mathematical expressions respectively to improve the robustness of the model to different writing styles. Experimental validation was performed on the CROHME 2014/2016/2019 dataset. After introducing the multi-scale feature fusion mechanism, the expression correctness rate reaches 55.25%, 52.31%, 53.72%, respectively. After introducing the mutual learning mechanism, the expression correct rate reaches 55.43%, 53.53%, 53.79%, respectively. The expression correctness rate reaches 58.88%, 55.10%, 57.05% after introducing both mechanisms at the same time. It is proved experimentally that the proposed method can effectively extract the features in formulas at different scales and overcome the problems of different handwriting styles and small amount of data by mutual learning mechanism. In addition, the experimental results on the HME100K dataset verified the effectiveness of the proposed model.

    CAI Xiaodong, ZHOU Qingsong, YE Qing
    2024, 52(2):  32-41.  doi:10.12141/j.issn.1000-565X.230023
    Asbtract ( 61 )   HTML ( 0)   PDF (1672KB) ( 22 )  
    Figures and Tables | References | Related Articles | Metrics

    The social recommendation model based on graph neural network has achieved good performance in improving the performance of the recommendation system. However, the existing methods ignored the possible feature mismatch between the queried target users and content nodes and their neighbors, which leads to the introduction of noise and reduces the model performance. To solve this problem, this paper proposed a social recommendation model DNSSR. Firstly, it constructed a relational graph containing multiple relationships between users and items, with richer information associations between nodes in the graph. Then the dynamic neighborhood sampling mechanism was used to obtain neighbor nodes that are more consistent with the characteristics of the target query pair, reducing noise information. In addition, in order to further improve the predictive performance of the model, this paper designed an enhanced graph neural network to model the sampled relationship subgraphs. It can distinguish the importance of different neighboring nodes and select more reliable information sources to obtain more robust user and item embedding vectors for rating prediction. The experimental results show that the prediction error of this model is significantly reduced compared to that of other advanced models, proving the effectiveness of the methods proposed in the paper. Especially for the dynamic neighborhood sampling mechanism, if it is abandoned, the RMSE and MAE indicators of DNSSR on the Ciao dataset will increase by 6.05% and 7.31% respectively, and the Epinions dataset will increase by 3.49% and 5.41% respectively, which fully demonstrate their effectiveness in reducing noise interference and improving the performance of social recommendation models.

    ZHOU Lang, FAN Kun, QU Hua, et al
    2024, 52(2):  42-49.  doi:10.12141/j.issn.1000-565X.230046
    Asbtract ( 1002 )   HTML ( 2)   PDF (1918KB) ( 88 )  
    Figures and Tables | References | Related Articles | Metrics

    The occurrence of fires has brought huge losses to society. The task of forest fire prevention and control is becoming increasingly urgent, and global warming has made this problem more complicated. Deep learning plays an important role in all walks of life. A large number of models are constantly designed and proposed, and there are various ways to improve the models. Therefore, this article proposed the EfficientNet-E model, which uses the more advanced ECA module (Efficient Channel Attention module) to replace the SE module in the EfficientNet. It improves the performance of the model by enhancing the performance of the attention mechanism. Compared with the SE module, the ECA module better retains the information during transmission, allowing the data features to be more fully retained during the transmission process, thus enabling the model to be optimized. To verify the performance of the EfficientNet-E model and the advantages of EfficientNet’s design idea in forest fire identification compared with traditional models, this article selected representatives of classic models, ResNet and DenseNet, as comparison references, and conducted related experiments in combination with EfficientNet and EfficientNet-E.The experiment selected 3 303 forest fire, non-fire and smoke pictures.The results of multiple rounds of tests show that EfficientNet-E is better than the conventional classic deep learning model in identifying forest fire data, and compared with the original EfficientNet’s average accuracy of 89.28%, EfficientNet-E’s average accuracy (90.04%) is obviously improved. The standard deviation is smaller and the training stability is better, which confirms the improved EfficientNet-E’s better performance.

    Energy,Power & Electrical Engineering
    LI Zhi, LIU Mingbo, LIU Hui, et al
    2024, 52(2):  50-61.  doi:10.12141/j.issn.1000-565X.220635
    Asbtract ( 74 )   HTML ( 0)   PDF (4592KB) ( 13 )  
    Figures and Tables | References | Related Articles | Metrics

    Virtual synchronous generator (VSG) simulates the operation mechanism of synchronous generator and introduces active frequency modulation and reactive voltage regulation control link. Different from the traditional current control converter, it is a voltage control strategy. Based on control configuration and power grid and source coordination characteristics, this paper constructed small signal analysis models, analyzed the oscillation modes under variable control parameters and power system parameters. Then it obtained high frequency resonance characteristics of voltage controlled VSG and proposed suppression control strategy. Firstly, this paper established the small-signal analysis models of voltage control virtual synchronous generator and traditional flow converter control and electrical link, respectively. Through eigenvalue analysis method, the oscillation mode and damping ratio of all the eigenvalues were quantitatively analyzed, and it is found that the voltage control virtual synchronous generator has a high frequency oscillation mode similar to the traditional converter. And it analyzed the control parameter change of the virtual synchronous generator and the high frequency resonance damping mechanism of the power grid scene. Then, based on the control strategy of voltage controlled virtual synchronous generator, with the active damping method, the virtual impedance control strategy based on the current inner ring front channel was proposed to suppress the high-frequency resonance. Finally, the proposed control strategy was verified in the loop experiment results using Matlab/Simulink simulation and RT-LAB controller. The results show that the virtual impedance control link effectively plays a positive damping role. Compared with the passive damping method, in the case of no additional measuring point and external control amount, the voltage control VSG network side damping characteristics are improved, the transient response ability is enhanced, and the high frequency instability phenomenon is effectively suppressed.

    ZHU Zhipeng, TANG Yong, SUN Yunwei, et al
    2024, 52(2):  62-73.  doi:10.12141/j.issn.1000-565X.230206
    Asbtract ( 89 )   HTML ( 0)   PDF (4845KB) ( 38 )  
    Figures and Tables | References | Related Articles | Metrics

    The swash plate piston pump is widely used in the hydraulic system of helicopter fatigue test, because its flow characteristics are critical to the stability of the hydraulic system. This paper used the three-dimensional geometric method to analyze the movement track of the plunger in the swash plate piston pump, and derived the formula of the overflow area based on the graphic analysis method. The corresponding Matlab program was compiled to realize the automatic calculation of the overflow area of oil suction and drainage stage.It constructed a single piston flow model considering flow reversal and leakage, combining AMESim software realized the transformation of the three-dimensional calculation model of the piston pump to a one-dimensional calculation model. Based on the calculation results of the initial operating conditions, this paper analyzed the flow characteristics of the piston pump and the reasons for its flow pulsation. The accuracy of the one-dimensional computational model was verified by the flow pulsation calculation results of the three-dimensional model based on the Computational Fluid Dynamics (CFD) method. On this basis, it analyzed the influence of different operating conditions (including operating temperature, working pressure and piston pump regulating parameters) on the outlet flow pulsation rate. And it analyzed the key points of adjusting and controlling the piston pump in helicopter fatigue test from the angle of reducing the flow pulsation rate and improving the test accuracy. The one-dimensional calculation model proposed in the paper can significantly improve the simulation calculation speed of piston pump, facilitate the improved design of the subsequent piston pump and be added to the virtual digital test platform as a subsystem.

    WU Chunling, FU Juncheng, XU Xianfeng, et al
    2024, 52(2):  74-83.  doi:10.12141/j.issn.1000-565X.220636
    Asbtract ( 129 )   HTML ( 0)   PDF (3784KB) ( 45 )  
    Figures and Tables | References | Related Articles | Metrics

    The existing State of Charge (SOC) estimation methods assume constant parameters for the battery model and do not consider the dynamic changes in these parameters, resulting in imprecise SOC estimates. In view of this limitation, the paper introduced an algorithm that combines online identification of battery model parameters with SOC estimation. Based on the second-order RC equivalent circuit model, it used Multi Innovation Least Squares (MILS) algorithm to identify the parameters in the lithium-ion battery model online, so as to modify the battery model in real time. At the same time, based on the modified battery model, it estimated the battery state of charge through Multi Innovation Extended Kalman Filter (MIEKF) algorithm. MILS algorithm can solve the problem of initial error accumulation in the process of online parameter identification, and can realize online accurate identification of model parameters. MIEK algorithm combines multi-innovation theory and Kalman filter theory, adds forgetting factor to weaken historical data and correct weight, solves the problem of data oversaturation, and has high accuracy and convergence. The experimental results show that, when identifying the parameters of the battery model, the Root Mean Square Error (RMSE) of the MILS algorithm is 1.4 mV, the RMSE of the RLS algorithm is 1.9 mV, and the estimation accuracy is improved by 26.3%. For the SOC estimation after parameter identification, the RMSE estimated by the MIEKF algorithm is 0.003 7, while the RMSE estimated by the EKF and AEKF algorithms are 0.007 3 and 0.005 2, respectively. The MIEKF algorithm improves the estimation accuracy by 49.31% compared to the EKF algorithm and by 28.84% compared to the AEKF algorithm. Moreover, in the case of an incorrect initial SOC value, the proposed algorithm can converge to the true value after about 30 seconds of battery operation. The algorithm proposed in the paper is an effective estimation method with high accuracy and good robustness.

    CHEN Guohua, XIE Mulin, ZHANG Qiang, et al
    2024, 52(2):  84-94.  doi:10.12141/j.issn.1000-565X.220422
    Asbtract ( 656 )   HTML ( 1)   PDF (3426KB) ( 33 )  
    Figures and Tables | References | Related Articles | Metrics

    Hydrogen vehicles have developed rapidly in recent years. Compared with hydrogen cars, hydrogen buses have greater safety risks due to their large sizes and large passenger capacity, so it’s of great significance to study the hydrogen leakage-explosion accidents and risks of hydrogen buses in station. The study predicted the dispersion of hydrogen under different leakage conditions after the failure of critical components of hydrogen bus with software FLACS. It studied the overpressure characteristics of the explosion after the explosion induced by the vapor cloud under different ignition locations according to the volume concentration distribution characteristics of hydrogen. And it analyzed the impact range of the accident, quantified the consequences of the explosion accident, and put forward the explosion risk prevention and control suggestions. Based on the risk analysis of hydrogen leakage in the parking lot, it put forward a risk analysis method of hydrogen leakage and explosions.The results show that the environmental wind and the planning layout of the station will affect the hydrogen dispersion after leakage. When the environmental wind direction is perpendicular to the length of the parking lot ceiling, it is most conducive to the dispersion of hydrogen; the dispersed velocity after the hydrogen leakage in the large sides station is fast; the hydrogen volume concentration decreases rapidly; the overpressure generated by ignition induced gas cloud explosion is less powerful; the maximum overpressure in the most severe leakage scenario is approximately 12.38 kPa. The high sensitivity hydrogen sensors can be installed near the TPRD device of hydrogen bus and at the corresponding ceiling position in the middle of the parking space to improve the sensitivity of real-time monitoring of hydrogen leakage.The risk of gas cloud explosion accident after hydrogen leakage in the parking lot is approximately 3.64×10-7 times per year, which is lower than the risk acceptable level, indicating the bus parking lot meets the safety requirements. The research results provides reasonable suggestions for the layout of hydrogen sensors in hydrogen buses and station, as well as the planning and construction of station.

    Green & Intelligent Transportation
    PENG Zixuan, CUI Lin, GUO Zhiwei, et al
    2024, 52(2):  95-103.  doi:10.12141/j.issn.1000-565X.230158
    Asbtract ( 250 )   HTML ( 0)   PDF (2077KB) ( 45 )  
    Figures and Tables | References | Related Articles | Metrics

    In view of the problem that passengers motivate ride-hailing drivers in the form of red packet or dispatching fees to realize self-scheduling, this paper studied the interactive relationship between passenger-ride-hailing matching decision and incentive strategy choice. Based on the matching equilibrium in taxi-sharing, this paper designed a matching equilibrium model for taxi-sharing with peer-passenger incentive mechanism, taking the maximization of total passenger surplus as the goal and considering the constraints such as matching, equilibrium and cost. From the perspective of passengers, it designed the passenger incentive strategy, ride-hailing incentive strategy, passenger and ride-hailing incentive strategy, and the three incentive strategies were embedded into the column generation algorithm to solve the model and to achieve matching equilibrium and pricing equilibrium. By empirical analysis of Dalian taxi data, the results show that, compared with with only motivating drivers from the supply side in the form of random dispatching fees, the implementation of incentive strategies from the demand side and the supply side can promote taxi-share, and the passenger surplus can be increased by 12.6%. When the demand is larger than the supply, about 26% of incentive is transferred among peer passengers for more taxi-share matches. The fare discount rate also affects the flow of incentives. Using incentive strategies and discount strategies simultaneously can avoid malicious competition and ineffective incentives. By increasing the number of taxi-sharing trips, we can simultaneously reduce travel costs for passengers and boost driver incomes.

    FENG Suwei, LIN Chang
    2024, 52(2):  104-112.  doi:10.12141/j.issn.1000-565X.230162
    Asbtract ( 186 )   HTML ( 0)   PDF (1647KB) ( 22 )  
    Figures and Tables | References | Related Articles | Metrics

    The algorithmic personalized pricing of online taxi-hailing platforms has produced complex market impacts, and compared with traditional cab service, the order cancellation rate of online taxi passengers reaches about 30%. Therefore, it is worth exploring the impact mechanism of algorithmic personalized pricing on passenger cancellation rates and the key characteristics of whether passengers fulfill their orders. This paper tried to establish the causal mechanism between algorithmic personalized pricing and passengers’ order cancellation rate using rectangular Hotelling model. Using a Stackelberg game model between two taxi-hailing platforms, it revealed the relationship among discriminatory pricing, passenger cancellation rate, and competition intensity between two platforms. Furthermore, based on the big data of online taxi-hailing platform orders, this paper applied some inductive learning tools such as Bhattacharyya distance, Gradient Boosting Decision Tree (GBDT) and improved Las Vegas method for wrapper-method feature selection to data mining of millions of orders on online taxi-hailing platforms to find out the key features that determine whether passengers take the orders or not. Analysis results show that the final consumption choice of passengers mainly depends on the price factors. And improving the match and dispatch strategies to reduce passengers’ waiting time can significantly improve fulfillment rate. The results are helpful for taxi-hailing platform to appropriately design the pricing and operation strategies to maintain the number of customers in the two-sided markets, which ensures the sustainable and successful operation of the platform. Meanwhile, it will provide a theoretical basis for antitrust authorities to intervene in platform personalized pricing.

    YANG Yazao, WENG Tangzheng
    2024, 52(2):  113-123.  doi:10.12141/j.issn.1000-565X.230228
    Asbtract ( 112 )   HTML ( 0)   PDF (2064KB) ( 18 )  
    Figures and Tables | References | Related Articles | Metrics

    As a self-driven travel mode, active travel or active mobility (AT or AM) plays a crucial role in alleviating urban traffic pressure and improving the temporal and spatial irrationality of urban traffic because of its small road occupation area, high mobility and good sustainability. The paper firstly studied the definition and connotation of active mobility. Secondly, starting from the spatiotemporal conditions affecting active mobility, it splited time and space into four types of constraints, namely, departure time, travel time, travel distance and built environment from the perspective of point and line, and expounded the characteristics of different types of spatiotemporal conditions in turn. It also analyzed the potential connection between active mobility and spatiotemporal constraints, and analyzed constraints influence on active mobility with single constraints and multiple spatiotemporal combinations as the entry point. Through summarization, it is found that the influence of spatiotemporal constraints on active mobility is more complicated than that in traditional cognition, and it is mainly a non-linear relationship and some of indexes have corresponding theoretical threshold. Finally, it put forward the shortcomings of the current theoretical research and the future research direction, which will provide a reference for the subsequent research. The purpose of this paper is to summarize the role of spatiotemporal constraints on active mobility, to show the specific change rule of movement travel under the role of different spatiotemporal constraints, aiming at changing the intrinsic thinking of travelers, improving the sharing rate of movement travel, and optimizing the travel structure of urban transportation.

    CAO Ningbo, ZHAO Liying
    2024, 52(2):  124-135.  doi:10.12141/j.issn.1000-565X.220489
    Asbtract ( 70 )   HTML ( 2)   PDF (2268KB) ( 15 )  
    Figures and Tables | References | Related Articles | Metrics

    In order to improve the management level of autonomous vehicles and pedestrians at intersections, and thus improve the operational efficiency and stability of traffic flow, the paper constructed a management method for autonomous vehicles and pedestrians at intersections based on maximum pressure control and autonomous vehicle trajectory planning methods. Firstly, the queue length of pedestrians was modeled by the probability distribution function and comprehensively considering the influence of arrival rate, crosswalk length and width, waiting time and arrival distribution on pedestrians. Based on the estimated pedestrian queue length, the maximum pressure control was adopted to develop a queuing length management method for autonomous vehicles and pedestrians at intersections. Then, in order to help the internal autonomous vehicles at the intersection avoid collisions and obtain the best movement trajectory, the maximum pressure control method and the trajectory planning method of the existing intersection were planned on the basis of controlling the queue length of the self-driving vehicles and pedestrians at the intersection. Finally, Python and SUMO, which is an open source traffic simulation software, were used to verify the model. The simulation lasts for 2 hours. The simulation results show that the proposed autonomous car and pedestrian management method can not only control the trajectory of autonomous vehicles, but also quickly stabilize and reduce their delay and queuing lengths, and improve the efficiency of intersection operation.

    JI Zhe, WANG Xin, YIN Hang, et al
    2024, 52(2):  136-144.  doi:10.12141/j.issn.1000-565X.230197
    Asbtract ( 152 )   HTML ( 5)   PDF (2921KB) ( 30 )  
    Figures and Tables | References | Related Articles | Metrics

    Selective catalytic reduction technology (SCR) is one of the commonly used technologies to reduce nitrogen oxide (NO x ) emissions from heavy-duty diesel vehicles. The efficiency of NO x conversion in the SCR system is closely related to the exhaust temperature. However, the existing NO x emission models primarily focus on vehicle driving conditions, neglecting the correlation with exhaust temperature. Thus, it increases the uncertainty of NO x emission measurement results, and challenges the establishment of emission inventory and the assessment of emission reduction policies. This study established a NO x emission rate library and a model based on actual vehicle operating conditions and measured emission data. Subsequently, an exhaust temperature model utilizing vehicle specific power (VSP) and heat loss coefficient was developed. Based on this, based on the chemical reaction principle in the SCR system, a NO x emission model incorporating exhaust temperature was derived. Finally, the proposed NO x model and the MOVES model (MOtor Vehicle Emission Simulator) were employed to estimate NO x emissions, which are then compared and analyzed against actual emissions. Results demonstrate the effectiveness of the proposed NO x emission model in real-world conditions, with relative errors of 9.1%, 3.9%, and 3.3% observed across three heavy-duty diesel buses. These errors represent a reduction of 24.0, 13.1, and 16.3 percentage points, respectively, when compared to the MOVES model. Additionally, analysis of NO x emission characteristics under different operating conditions reveals that the average NO x conversion rate of heavy diesel trucks is 39.2 percentage points higher than that of diesel buses.

News
 
Featured Article
Most Read
Most Download
Most Cited