GWGuilin Wen

Guilin Wen

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Guilin Wen has a total of 54 co-authors. Here are the top co-authors by number of shared publications:

  • Jie Liu

    Center for Research on Leading Technology of Special Equipment, School of Mechanical and Electric Engineering, Guangzhou University, Guangzhou, People’s Republic of China

    6 shared publications
  • Hanfeng Yin

    State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body, Hunan University, Changsha, People’s Republic of China

    5 shared publications
  • Qixiang Qing

    State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body

    4 shared publications
  • Huidong Xu

    Taiyuan Univ Technol, Coll Mech, Taiyuan 030024, Shanxi, Peoples R China

    3 shared publications
  • Jie Liu

    Hebei Innovation Center for Equipment Lightweight Design and Manufacturing, School of Mechanical Engineering, Yanshan University, Qinhuangdao, China

    3 shared publications
  • Shan Yin

    Hunan University

    3 shared publications

Latest Publications

  • Open Access

    Study on the Microscopic Mechanism of Axle Steel EA4T during Uniaxial Cyclic Deformation Process

    by Xuehong Ren, Shaopu Yang, Wenjie Zhao, Guilin Wen

    In order to reveal the dislocation evolution law of body-centered cubic axle steel EA4T during cyclic deformation and provide an experimental basis for the subsequent construction of cyclic constitutive models based on microscopic physical mechanisms, macroscopic deformation experiments were first conducted on axle steel EA4T, including monotonic tensile experiments under different deformation amounts, symmetric strain cycling experiments under a different number of cycles, and ratcheting deformation experiments under a different number of cycles. Then, systematic observations of different samples at different deformation stages were conducted using a Transmission Electron Microscope (TEM) to investigate the dislocation configuration and evolution during strain cycling and ratcheting deformation. The observed results show that the dislocation evolution law of axle steel EA4T during the uniaxial tensile experiment, symmetrical strain cycling, and ratcheting deformation is basically the same, and the dislocation density increases with the increase in plastic deformation and number of cycles. The dislocation configuration gradually develops from low-density dislocation configurations such as dislocation lines and dislocation pileups to high-density dislocation configurations such as severe dislocation tangles and dislocation walls. The microscopic mechanism of the uniaxial ratcheting evolution of axle steel EA4T can be qualitatively explained by the dislocation configuration and evolution.


  • Open Access

    Deep Subdomain Transfer Learning with Spatial Attention ConvLSTM Network for Fault Diagnosis of Wheelset Bearing in High-Speed Trains

    by Jiujian Wang, Shaopu Yang, Yongqiang Liu, Guilin Wen

    High-speed trains operate under varying conditions, leading to different distributions of vibration data collected from the wheel bearings. To detect bearing faults in situations where the source and target domains exhibit differing data distributions, the technique of transfer learning can be applied to move the distribution of features gleaned from unlabeled data in the source domain. However, traditional deep transfer learning techniques do not take into account the relationships between subdomains within the same class of different domains, resulting in suboptimal transfer learning performance and limiting the use of intelligent fault diagnosis for wheel bearings under various conditions. In order to tackle this problem, we have developed the Deep Subdomain Transfer Learning Network (DSTLN). This innovative approach transfers the distribution of features by harmonizing the subdomain distributions of layer activations specific to each domain through the implementation of the Local Maximum Mean Discrepancy (LMMD) method. The DSTLN consists of three modules: a feature extractor, fault category recognition, and domain adaptation. The feature extractor is constructed using a newly proposed SA-ConvLSTM model and CNNs, which aim to automatically learn features. The fault category recognition module is a classifier that categorizes the samples based on the extracted features. The domain adaptation module includes an adversarial domain classifier and subdomain distribution discrepancy metrics, making the learned features domain-invariant across both the global domain and subdomains. Through 210 transfer fault diagnosis experiments with wheel bearing data under 15 different operating conditions, the proposed method demonstrates its effectiveness.


  • Open Access

    An immediate-return reinforcement learning for the atypical Markov decision processes

    by Zebang Pan, Guilin Wen, Zhao Tan, Shan Yin, Xiaoyan Hu

    The atypical Markov decision processes (MDPs) are decision-making for maximizing the immediate returns in only one state transition. Many complex dynamic problems can be regarded as the atypical MDPs, e.g., football trajectory control, approximations of the compound Poincaré maps, and parameter identification. However, existing deep reinforcement learning (RL) algorithms are designed to maximize long-term returns, causing a waste of computing resources when applied in the atypical MDPs. These existing algorithms are also limited by the estimation error of the value function, leading to a poor policy. To solve such limitations, this paper proposes an immediate-return algorithm for the atypical MDPs with continuous action space by designing an unbiased and low variance target Q-value and a simplified network framework. Then, two examples of atypical MDPs considering the uncertainty are presented to illustrate the performance of the proposed algorithm, i.e., passing the football to a moving player and chipping the football over the human wall. Compared with the existing deep RL algorithms, such as deep deterministic policy gradient and proximal policy optimization, the proposed algorithm shows significant advantages in learning efficiency, the effective rate of control, and computing resource usage.


  • Coupled CFD/MBD Method for a Tilt Tri-rotor UAV in Conversion of Flight Modes

    by Guilin Wen, Dong Wu, Hanfeng Yin, Daibing Zhang

    Aerodynamic characteristics of tilt-rotor UAVs are usually simulated by computational fluid dynamics (CFD) only, and the fuselage/wing is fixed in simulations. It is difficult for this method to consider drastic changes of the aerodynamic forces and flowfield originating from conversion of flight modes. And during flight-mode conversions, the rapid changes of postures of tilt-rotors will unavoidably bring great influence on aerodynamic characteristics. In addition, the simulation of the control system for flight-mode conversions cannot be realised by CFD only. In this paper, considering the motion of the fuselage/wing and the dynamic tilting process of tilt-rotors, a new computational method by coupling CFD and multi-body dynamics (MBD) is presented for flight-mode conversions of a tilt tri-rotor UAV with the large-space computational domain. The numerical results validate the feasibility of the coupling method and also show effects of the fuselage/wing motion and flight-mode conversions to aerodynamic characteristics of the tilt tri-rotor UAV.


  • An efficient evolutionary structural optimization method with smooth edges based on the game of building blocks

    by Hongxin Wang, Jie Liu, Guilin Wen

    Although bi-directional evolutionary structural optimization (BESO) has the advantages of a simple concept and clear-cut solutions, the edges of the optimal solution are normally non-smooth. This article proposes an approach that directly represents the smooth structure using finite elements. The pseudo-auxiliary line is introduced to make the staggered boundary approach a smooth one, and a rule is presented to make the boundary element deformable to describe the structure’s edges. To improve the efficiency, the game of building blocks is incorporated into the classical BESO algorithm. A set of basic blocks is predefined and is positioned in a suitable location to assemble the optimal structure. In this way, the optimal structure has better mechanical performance and smooth edges with acceptable computational cost. The roundness of corners is easily controlled by adjusting the configuration of the basic blocks. Several two- and three-dimensional numerical examples investigate the effects of the parameters of auxiliary lines and the shape of basic blocks on the boundary smoothness and optimization performance. The efficiency of the proposed method is justified through these examples.


  • An adaptive mesh‐adjustment strategy for continuum topology optimization to achieve manufacturable structural layout

    by Hongxin Wang, Jie Liu, Guilin Wen

    Summary: This paper presents an adaptive mesh adjustment algorithm for continuum topology optimization method to describe the structural boundary using nonuniform isoparametric element. A criterion on the basis of the node movement is proposed; herein, the densities and coordinates of the nodes are defined to instruct the deformation of finite elements in subsequent optimization iterations. With such a scheme, the topology optimization can start from a regular mesh discretization then gradually yields an optimal design with clear structural boundaries. The element in the transition along the boundary is refined; on the contrary, the pure solid or void element is coarsen. The contribution of this work is to improve the resolution of the structural boundaries and decrease the percentage of transitional regions with the invariant design variable. Several 2D and 3D numerical examples indicate the effectiveness of our proposed method. Seen from the examples, the structural boundary become smoother and the intermediate densities have been reduced up to 70%. In addition, a design process based on the presented method is proposed to make the optimum solutions be fabricated conveniently and accurately by linking it with the 3D design software, ie, SolidWorks, which is also demonstrated in the numerical examples.


  • Open Access

    An Efficient Topology Description Function Method Based on Modified Sigmoid Function

    by Xingfa Yang, Jie Liu, Yin Yang, Qixiang Qing, Guilin Wen

    Optimal geometries extracted from traditional element-based topology optimization outcomes usually have zigzag boundaries, leading to being difficult to fabricate. In this study, a fairly accurate and efficient topology description function method (TDFM) for topology optimization of linear elastic structures is developed. By employing the modified sigmoid function, a simple yet efficient strategy is presented to tackle the computational difficulties because of the nonsmoothness of Heaviside function in topology optimization problem. The optimization problem is to minimize the structural compliance, with highest stiffness, while satisfying the volume constraint. The design problem is solved by a Sequential Linear Programming method. Convergent, crisp, and smooth final layouts are obtained, which can be fabricated without postprocessing, demonstrated by a series of numerical examples. Further, the proposed method has a rather higher accuracy and efficiency compared with traditional TDFM, when the classical topology optimization methods, such as bidirectional evolutionary structural optimization (BESO) and solid isotropic material with penalization (SIMP) method, are taken as benchmark.


  • Open Access

    Mechanical behaviour of a creased thin strip

    by Jie Liu, Shanqing Xu, Guilin Wen, Yi Min Xie

    In this study the mechanical behaviour of a creased thin strip under opposite-sense bending was investigated. It was found that a simple crease, which led to the increase of the second moment of area, could significantly alter the overall mechanical behaviour of a thin strip, for example the peak moment could be increased by 100 times. The crease was treated as a cylindrical segment of a small radius. Parametric studies demonstrated that the geometry of the strip could strongly influence its flexural behaviour. We showed that the uniform thickness and the radius of the creased segment had the greatest and the least influence on the mechanical behaviour, respectively. We further revealed that material properties could dramatically affect the overall mechanical behaviour of the creased strip by gradually changing the material from being linear elastic to elastic-perfect plastic. After the formation of the fold, the moment of the two ends of the strip differed considerably when the elasto-plastic materials were used, especially for materials with smaller tangent modulus in the plastic range. The deformation patterns of the thin strips from the finite element simulations were verified by physical models made of thin metal strips. The findings from this study provide useful information for designing origami structures for engineering applications using creased thin strips.


  • On the ensemble of metamodels with multiple regional optimized weight factors

    by Hanfeng Yin, Hongbing Fang, Guilin Wen, Matthew Gutowski, Youye Xiao