Table of Content

15 August 2021, Volume 38 Issue 4 Previous Issue    Next Issue

Select

Deep Learning Based 2D Face Recognition: a Survey YU Cui-can, LI Hui-bin 2021, 38 (4):  451-469.  doi: 10.3969/j.issn.1005-3085.2021.04.001 Abstract ( 2323 )   PDF (2019KB) ( 1635 )   Save Compared with iris, fingerprint, gait, and other biometric recognition technologies, face recognition has attracted wide attention from academia to industry due to its unique advantages such as natural, convenient, and user-friendly experience. In recent years, driven by deep learning technology, face recognition has made a breakthrough, which shows strong robustness even when suffering from obstacles like facial expression, head pose, illumination, and external occlusions. In particular, deep face recognition technologies have been widely used in security, finance, education, transportation, new retail, and other applications. We realise that in the process of deep face recognition technology becoming widespread, there is an urgent need for some review articles to summarise the basic principles and methods of deep face recognition. This paper first briefly reviews the development of face recognition and then introduces the deep learning based face recognition methods from five aspects: face preprocessing, deep feature learning, feature comparison, face datasets, and evaluation. Finally, the development trend of deep face recognition is discussed. Related Articles | Metrics

Select

Research on Energy-saving Optimized Control Technology of Urban Rail Transit Trains Based on Intelligent Computing ZHANG Fang, CUI Wei-chen, GAO Li-min, DONG Chun-zhao, ZHENG Jing-wen 2021, 38 (4):  470-482.  doi: 10.3969/j.issn.1005-3085.2021.04.002 Abstract ( 306 )   PDF (351KB) ( 335 )   Save With the rapid growth of urban rail transit mileage, the energy-saving optimised control technology of urban rail trains is a hot topic in recent years. The optimised operation of energy-saving is of great economic significance and social benefit to the actual operation of urban rail transit. At present, the research on this issue mainly relies on field trials and experts experience. In this paper, a nonlinearly constrained optimisation scheme based on the combination of cruising speed and cruising distance is proposed for an actual urban rail transit line. The particle swarm optimisation and genetic algorithm in the intelligent computing field are used to solve the energy-saving optimisation operation curve. The experimental results show that the results obtained by particle swarm optimisation and genetic algorithm can guide train operation. Related Articles | Metrics

Select

Fault Detection Methods for Catenary Dropper of High-speed Railway ZHANG Xue-wu 2021, 38 (4):  483-489.  doi: 10.3969/j.issn.1005-3085.2021.04.003 Abstract ( 1001 )   PDF (1141KB) ( 214 )   Save The droppers are the main components of the high-speed railway catenary system. They are prone to fracture and relaxation, which directly threatens the safety of driving. Under the action of fluctuating wind and pantograph simultaneously, the acceleration signals obtained by the acceleration sensors installed on the messenger wire and contact line are relatively strong. Using the LSTM network model, it is easy to predict the fracture and relaxation of the droppers. In this paper, the strong feature extraction ability of the convolutional neural network and the time sequence expression ability of recurrent neural network are used to solve the problem of the detection of dropper fault under the action of fluctuating wind only. At the same time, the attention mechanism is introduced to establish the fusion network model of CNN-LSTM-Attention. In the process of network training, the Bayesian optimisation method is used to select hyperparameter. Compared with the LSTM model, the fusion network model improves the detection accuracy of dropper fault and has strong practicability. Related Articles | Metrics

Select

Reduction of Multi-decision Covering Information Systems XU Li-ting, LI Jin-jin, LIN Yi-dong, LI Yi-liang 2021, 38 (4):  490-502.  doi: 10.3969/j.issn.1005-3085.2021.04.004 Abstract ( 204 )   PDF (206KB) ( 201 )   Save The covering decision information system is one of the research focuses. However, there is little research on the covering decision information system which the decision is a covering. In this paper, a characteristic function is used to transform the decision expressed by covering into a formal context composed of 0 and 1. Furthermore, the definition of a multi-decision covering information system is given. In the multi-decision covering information system, two kinds of covering reduction are discussed by constructing a corresponding discernibility matrix. And the relation between the covering consistent set corresponding to these two kinds of reductions and the upper approximation consistent set in the covering decision information system is studied. Related Articles | Metrics

Select

Application of Ant Colony Algorithm in Intelligent LED Street Lamp Control LV Ai-hua 2021, 38 (4):  503-512.  doi: 10.3969/j.issn.1005-3085.2021.04.005 Abstract ( 660 )   PDF (482KB) ( 310 )   Save To optimise the LED street lamp and achieve effective monitoring, reduce node energy consumption, and improve the efficiency of network node processing and data dissemination in LED street lamp control, an addressing method for LED street lamp low-voltage distribution network is developed based on an ant colony algorithm, to improve the algorithm's strength of optimal path search. For the phenomenon of low calculation accuracy and high energy consumption, an improved measure of the ant colony algorithm is designed to avoid the algorithm falling into a local optimum. According to the performance of PLC managing street lamps, the algorithm constructs the network topology model. Based on the optimisation objective function and state transition law, the algorithm updates pheromone, revises and improves elements around the function objective. Finally, from state transition law and search, the algorithm designs automatic streetlight routing model and draws up automatic routing protocol framework. With the low-voltage distribution network, the channel quality of the model achieves dynamic identification. When the channel quality changes, the communication carrier of the power line network completes dynamic routing maintenance to realise the reliability of the communication network. The algorithm reduces the network energy consumption, effectively avoids network congestion, improves the average hit rate of the street light route, and reduces the route execution time. Simulation results verify the reliability and effectiveness of the algorithm. Related Articles | Metrics

Select

Arbitrary Slow Convergence of Shannon Sampling Reconstruction with Oversampling Technique WU Ying, GAO Meng-yao, ZHANG Xue-lin 2021, 38 (4):  513-521.  doi: 10.3969/j.issn.1005-3085.2021.04.006 Abstract ( 236 )   PDF (195KB) ( 188 )   Save Shannon sampling theorem is an important conclusion in signal processing. It states that the exact reconstruction of a bandlimited signal can be obtained from its samples at Nyquist rate by the cardinal series. It is of theoretical interest to study the convergence rate of the reconstruction of Shannon sampling. In this work, the convergence rate of Shannon's reconstruction is analysed by using the theory on the slow convergence of operator sequences. It is provided that Shannon's reconstruction consists of a sequence of ``arbitrarily slow" convergent operators. Specifically, for any positive sequence $\alpha(n)\to 0$, there exists a bandlimited signal $f$ such that the $n$-th truncation error of its cardinal series is larger than $\alpha(n)$ for all $n$, where the truncation errors are measured in $L^p$ norms, for $1<p<\infty$. It is also shown that the acceleration techniques by over-sampling and convergence factor do not resolve the slowness, which is still ``arbitrarily slowly". Related Articles | Metrics

Select

A Bootstrap Test for the Varying Coefficient Spatial Autoregressive Models DU Ying, LI Ti-zheng 2021, 38 (4):  539-552.  doi: 10.3969/j.issn.1005-3085.2021.04.008 Abstract ( 451 )   PDF (182KB) ( 497 )   Save The varying coefficient spatial autoregressive model is a generalisation of a varying coefficient model in spatial data analysis. It has been widely valued and studied because of its many application backgrounds. To apply the models, the first problem is to confirm whether the coefficients in the model change with respect to a variable. Based on the Bootstrap test, the identification of constant-coefficient terms in the varying coefficient spatial autoregressive models is studied, which provides a basis for the confirmation of a semi-varying coefficient spatial autoregressive model. Furthermore, some simulation experiments are conducted to evaluate the validity of the Bootstrap approximation in the case of a finite sample size. Meanwhile, when the error term distribution is different. The value of the spatial autoregressive parameter changes, and the collinearity among the explanatory variables varies, the accuracy of the bootstrap approximation to their null distributions and the power of the test are investigated. The simulation results demonstrate that the proposed Bootstrap method can accurately approximate the zero distribution of test statistics, and the test has a good effect. Related Articles | Metrics

Select

Two Time-mesh Finite Element Method for Cahn-Hilliard Equation WANG Dan-xia, JIA Hong-en, LI Ya-qian 2021, 38 (4):  553-563.  doi: 10.3969/j.issn.1005-3085.2021.04.009 Abstract ( 317 )   PDF (229KB) ( 343 )   Save A time two-mesh (TT-M) finite element (FE) method is proposed for solving the Cahn-Hilliard equation in a nonlinear numerical scheme. The method is carried out in two steps. A nonlinear Cahn-Hilliard system is solved on time coarse mesh at the first step, where the finite element method is used for spatial discretisation, and the Crank-Nicolson scheme is used for time discretisation. The second step is that a linear problem is solved on time fine mesh. Finally, the stability analysis and error estimates of the proposed method is given. Numerical examples are given to confirm the theoretical analysis. The results show that the method of this paper can save computation time compared with the traditional Galerkin finite element method. The validity and feasibility of the proposed method are illustrated. Related Articles | Metrics

Select

The Generalized Solution for the Nonlinear Singular Perturbation Robin Problem of Integral-differential Evolution Equation FENG Yi-hu, MO Jia-qi 2021, 38 (4):  564-572.  doi: 10.3969/j.issn.1005-3085.2021.04.010 Abstract ( 338 )   PDF (333KB) ( 211 )   Save A class of generalised nonlinear singular perturbation integral-differential evolution equation Robin problem is discussed. Firstly, the outer solution of the model is obtained by using a solving method of the Fredholm type integral equation. Next, the boundary layer corrective term of the solution is constructed by using the variables of multiple scales. Then the initial layer corrective term of the solution for the original model is found using the stretched variable. And the composed expansion of formal solution for the singular perturbation problem is constructed. Finely, the asymptotic expansion of the generalised solution is proved by using the fixed point theory of the functional analysis. Related Articles | Metrics

Select

Influence of Human Behavior on Quitting Smoking Dynamics LI Zhi-min, GAO Jian-zhong, ZHANG Tai-lei, FANG Shu, SONG Xue-li 2021, 38 (4):  573-585.  doi: 10.3969/j.issn.1005-3085.2021.04.011 Abstract ( 282 )   PDF (177KB) ( 520 )   Save This paper studies the impact of human behavior on quitting smoking dynamics by a PLSQ mathematical model. We start with a quitting smoking ordinary differential equation (ODE) model with nonlinear incidence and human behavior impacts. The basic reproduction number of the model is obtained by establishing the basic reproduction matrix. Using the Routh-Hurwitz criterion and Lyapunov functionals and LaSalle's invariant principle and the second additive compound matrix, local and global dynamics of the model are analysed. Based on the partial rank correlation coefficients (PRCCs), the influence of various parameters in the basic reproduction number on the transmission of smokers is analysed. Particularly, the influence of human behavior on the transmission of smokers is obtained through theoretical deduction. Finally, the theoretical derivation is verified by numerical simulation. Our results show that human behavior can reduce the number of smokers. Related Articles | Metrics

Select

Pattern Dynamics of Vegetation System with Holling-type II and Nonlocal Delay LIANG Juan, LI Li, CUI Liang, GUO Zun-guang 2021, 38 (4):  586-600.  doi: 10.3969/j.issn.1005-3085.2021.04.012 Abstract ( 498 )   PDF (347KB) ( 523 )   Save In arid or semi-arid regions, vegetation absorbs water through the nonlocal effects of roots. The paper establishes a mathematical model with nonlocal delay and Holling-II functional response function. By mathematical analysis, the conditions under which the vegetation-water model generates the Turing pattern are obtained. The spatial distributions of vegetation under different delays are obtained by numerical simulation. The simulation results show that delay in vegetation density presents a ``parabolic phenomenon", and delay can cause a change in the pattern structure. Specifically, as delay increases, a pattern is transformed from uniform distribution to nonuniform distribution. When the delay parameter is less than the threshold, vegetation density increases with the decrease of delay. On the contrary, the density of vegetation will increase with the increase of delay. Besides, the functional response term coefficient is positively correlated with vegetation density. The numerical simulation results show the influence of nonlocal delay and Holling-II functional response function on vegetation pattern, which provide a new theoretical basis for vegetation protection. Related Articles | Metrics