首页期刊导航|Automatic Control and Computer Sciences
期刊信息/Journal information
Automatic Control and Computer Sciences
Allerton Press
Automatic Control and Computer Sciences

Allerton Press

0146-4116

Automatic Control and Computer Sciences/Journal Automatic Control and Computer SciencesSCIEIESCI
正式出版
收录年代

    Algorithms for the Analysis of Queueing System M/G/1/infinity with Cut-Off of the Line

    Dalinger, I. M.Andronov, A. M.
    14页
    查看更多>>摘要：Theclassical queueing system M/G/1 is considered for a case when the service begins, if the number of waited claims reaches the fixed level k. Various algorithms are considered for the calculation of such indices as the mean and the distribution of the waiting time, the queue length and so on.
      原文链接:
    • NSTL
    • Pleiades Publishing Inc

    Neural Translation System of Meta-Domain Transfer Based on Self-Ensemble and Self-Distillation

    Liu, YupengZhang, YananZhang, Xiaochen
    11页
    查看更多>>摘要：In order to address the shortcoming that feature representation limitation in machine translation (MT) system, this paper presents a transfer method for features in MT. Its main aim is to solve knowledge transfer of different training corpus in the decoding process. In this paper, the meta domain is modeled. A model agnostic self-ensemble and self-distillation training framework is proposed. The training is divided into model training and meta training to better adapt to the two types of features. In this paper, we have done extensive experiments on the classical neural machine translation system, and the model is compared with the classical methods. The experimental results show that the proposed model has improved in the transfer task of different domains and systems. In this paper, translation knowledge transfer is carried out on the Chinese-English translation dataset in the subdivided domain, which has a significant performance improvement in the news, education and law domain.
      原文链接:
    • NSTL
    • Pleiades Publishing Inc

    Time Series Similarity Search Methods for Sensor Data

    Jawale, AnupamaMagar, Ganesh
    10页
    查看更多>>摘要：Time series is a type of dynamic data used in many applications. Time series speed may vary from milliseconds to years or decades. In past decade, rise in various sensor based technologies have made time series sensor data available easily and in larger extent. Therefore, high dimensionality of the data in customized applications is always a challenging task for efficient mathematical computing accuracy and performance optimization. One of the major operations performed on time series is finding out similarity between two or more time series. Two time series can be considered similar on the basis of distance between them. Computation of these distances is achieved by various methods. This research study aims to compare eight such methods for accelerometer sensor data collected from smartphone based accelerometer during car and scooter ride. This study also proposes a modified method of distance computation considering tyre pressure and weight of the vehicle. Research findings have shown that modified method of DTW (dynamic time warping) is proved more efficient in distinguishing time series generated by two different weights' vehicles. Results have shown as maximum of 67% recognition rate is achieved by modified DTW method compared to traditional DTW method.
      原文链接:
    • NSTL
    • Pleiades Publishing Inc

    Hybrid Path Planning Algorithm of the Mobile Agent Based on Q-Learning

    Gao, TengtengLi, CaihongLiu, GuomingGuo, Na...
    13页
    查看更多>>摘要：In the path planning using Q-learning of the mobile agent, the convergence speed is too slow. So, based on Q-learning, two hybrid algorithms are proposed to improve the above problem in this paper. One algorithm is combining Manhattan distance and Q-learning (CMD-QL); the other one is combining flower pollination algorithm and Q-learning (CFPA-QL). In the former algorithm, the Q table is firstly initialized with Manhattan distance to enhance the learning efficiency of the initial stage of Q-learning; secondly, the selection strategy of the epsilon-greedy action is improved to balance the exploration-exploitation relationship of the mobile agent's actions. In the latter algorithm, the flower pollination algorithm is first used to initialize the Q table, so that Q-learning can obtain the necessary prior information which can improve the overall learning efficiency; secondly, the epsilon-greedy strategy under the minimum value of the exploration factor is adopted, which makes effective use of the action with high value. Both algorithms have been tested under known, partially known, and unknown environments, respectively. The test results show that the CMD-QL and CFPA-QL algorithms proposed in this paper can converge to the optimal path faster than the single Q-learning method, besides the CFPA-QL algorithm has the better efficiency.
      原文链接:
    • NSTL
    • Pleiades Publishing Inc

    An Evader Control Strategy in the Non-Linear Differential Game Problem with Terminal Limitations

    Shcherban, I. V.Ivanov, S. V.Shcherban, O. G.
    7页
    查看更多>>摘要：This article considers the method of developing an evader control strategy in the non-linear differential pursuit-evasion game problem. It is assumed that the pursuer resorts to the most probable control strategy in order to capture the evader and that at each moment the evader knows its own and the enemy's physical capabilities. This assumption allows to bring the game problem down to the problem of a unilateral evader control, with the condition of reaching a saddle point not obligatory to be fulfilled. The control is realised in the form of synthesis and additionally ensures that the requirements for bringing the evader to a specified area with terminal optimization of certain state variables are satisfied.
      原文链接:
    • NSTL
    • Pleiades Publishing Inc

    Visual Tracking Method Based on Siamese Network with Multi-Feature Fusion

    Li, QingdangXu, RuiZhang, MingyueSun, Zhen...
    10页
    查看更多>>摘要：The traditional deep learning tracking method SiamFC faces performance degradation while solving issues, for instance, similar background, occlusion, target deformation, and illumination variation. This paper proposes an improved SiamFC with multi-feature fusion strategy. The proposed method first extracts the histogram of gradient and color name of the template image and search area by correlation filter. Then, the method fuses them and weights the SiamFC response map to obtain a more accurate object response position. Comparison experiments on VOT and OTB datasets prove that the improved method is more accurate and robust than the excellent tracking methods to deal with problems such as target cover, out of sight, scale variation and motion blur.
      原文链接:
    • NSTL
    • Pleiades Publishing Inc

    Griffiths' Variable Learning Rate Online Sequential Learning Algorithm for Feed-Forward Neural Networks

    Bharath, Y. K.
    6页
    查看更多>>摘要：For online sequential training of deep neural networks, where the training data set is chaotic in nature, it becomes quite challenging for choosing a proper learning rate. This paper presents Griffiths' variable learning rate algorithm for improved performance of online sequential learning of feed-forward neural networks used for chaotic time-series prediction. Here, the learning rate is varied based on Griffiths' cross-correlation between input training data and squared error, which facilitates better tracking of time-series data.
      原文链接:
    • NSTL
    • Pleiades Publishing Inc

    Analysis of Dynamic Characteristics of Man-Machine Co-Driving Vehicle during Driving Right Switching

    Ge, Guangcheng
    14页
    查看更多>>摘要：In order to study the influence of different drivers and vehicle speeds on the dynamics of smart cars when the driving rights are switched; this paper analyzes them through driving experiments. This experiment recruited 16 participants, and built a virtual experimental platform for man-machine co-driving, and designed an experimental program at three speeds of 50, 80, and 120 km/h based on 8 s early warning time interval for driving rights to take over. And the experimental data is processed and analyzed by K-means clustering. The results show that the driver's age and driving experience affect the driver's take-over behavior and the dynamics of the vehicle. The take-over behavior taken by the high driving experience or young driver in the process of driving right switching can ensure that the vehicle has better stability. On the one hand, the increase in vehicle speed will affect the driver's take-over behavior. On the other hand, it will cause the nonlinear characteristic of the vehicle to be significant and the driving stability of the vehicle to be worse.
      原文链接:
    • NSTL
    • Pleiades Publishing Inc

    Recognition Confidence of Welding Seam Defects in TOFD Images Based on Artificial Intelligence

    Liu, QinWang, YiminYe, BirunDing, Min...
    9页
    查看更多>>摘要：In this paper, a target detection approach (Faster R-CNN) based on convolutional neural networks is applied to the training and recognition of typical defects in TOFD welding seam images. Before training and recognition, a total of 162 ultrasonic TOFD welding seam images containing five typical defects are collected. The ultrasonic TOFD welding seam image dataset required for neural network model training is established on the basis of the collected images. The neural network model is trained including pre-training on ImageNet, RPN training alone, Faster R-CNN training alone, and joint RPN and Faster R-CNN training. During training, the parameters in the program are adjusted, and, then, the convergence of the neural network models and recognition performance are compared after the same training iterations. It is found that the neural network model has a tendency to converge only when the batch size is 10 and the learning rate is 0.001. Under this parameter configuration condition, the program conducts training with more iterations and is used to identify welding seam defects. The results show that the program is accurate in locating typical defects in the images, and the recognition confidence for all kinds of defects is more than 0.9. Compared with the other parameter configuration conditions after the same training iterations, the program has the highest recognition confidence in identifying all types of defects when the batch size is 10 and the learning rate is 0.001.
      原文链接:
    • NSTL
    • Pleiades Publishing Inc