Automatic process modeling has become an effective tool in reducing the lead-time and the cost for designing forming processes. The numerical modeling process is performed on a fully coupled damage constitutive equations and the advanced 3D adaptive remeshing procedure. Based on continuum damage mechanics, an isotropic damage model coupled with the Johnson–Cook flow law is proposed to satisfy the thermodynamic and damage requirements in metals. The Lemaitre damage potential was chosen to control the damage evolution process and the effective configuration. These fully coupled constitutive equations have been implemented into a Dynamic Explicit finite element code Abaqus using user subroutine. On the other hand, an adaptive remeshing scheme in three dimensions is established to constantly update the deformed mesh to enable tracking of the large plastic deformations. The quantitative effects of coupled ductile damage and adaptive remeshing on the sheet metal forming are studied, and qualitative comparison with some available experimental data are given. As illustrated in the presented examples this overall strategy ensures a robust and efficient remeshing scheme for finite element simulation of sheet metal-forming processes.

An important step in the manufacturing processes of thin composite components is the layingup of the reinforcement onto the mould surface. The prediction of the angular distortion of the reinforcement during draping and the changes in fibre orientation are essential for the understanding of the manufacture process and the evaluation of the mechanical properties of the composite structures. This paper presents an optimization-based method for the simulation of the forming processes of woven fabric reinforced composites. Two different approaches are proposed for the simulation of the draping of woven fabric onto complex geometries: geometrical and mechanical approaches. The geometrical approach is based on a fishnet model. It is well adapted to predimensioning fabrics and to give a suitable quantification of the resulting flat patterns. The mechanical approach is based on a mesostructural model. It allows us to take into account the mechanical properties of fibres and resin and the various dominating mode of deformation of woven fabrics during the forming process. Some numerical simulations of the forming process are proposed and compared with the experimental results in order to demonstrate the efficiency of our approaches.

Matching manually cropped pedestrian images between queries and candidates, termed as person re-identification, has achieved significant progress with deep convolutional neural networks. Recently, a topic called ‘person search’ is proposed for the end-to-end application of re-identification technologies. It integrates object detection and person re-identification and aims to both locate and match pedestrians on a gallery of raw images. However, the design and implementation of such kind of hybrid network are difficult and computationally consuming in real practical situations. In order to fasten the design and ease the implementation, this paper proposes a deep-frozen transfer learning framework, named FT-MDnet, to extract re-identification features from a pre-trained detection network in two steps. First, using a channel-wise attention mechanism, a network called adaptive transfer learning network (ATLnet) is used to convert the sharing data of the underlying detection network to a re-identification feature map. Then, a multi-branch feature representation network called multiple descriptor network (MDnet) is proposed to extract re-identification features from the re-identification feature map. Our proposed solution has been verified on different types of mainstream detection networks, including YOLOv3, YOLOv4, Mask RCNN, and CenterNet. The experimental results show that our solution outperforms all other person search solutions by a large margin. It proves that the feature representations of detection networks are highly compatible with re-identification, and the proposed framework effectively extracts these features out. To encourage further research, we have made our framework open source.