No abstract

No abstract

No abstract

Purpose This paper aims to investigate the additive manufacturing (AM) approach of a spatial complex curve feature (SCCF, mapped from two-dimensional nonuniform rational B-splines [2D-NURBS] curve) on a complex surface based on a serial robot using plasma built-up welding, and lays a foundation for plasma AM SCCFs on complex surfaces by combining the NURBS theory with the serial robotic kinematics. Design/methodology/approach Combining serial robotic kinematics and NURBS theory, a SCCF mapped from a square-like 2D-NURBS curve is prepared on a predefined complex NURBS surface using serial robotic plasma AM. The interpolation points C (ui) on the square-like 2D-NURBS curve are obtained using the equi-chord length interpolation method, and mapped on a predefined NURBS surface to get mapped points S (ui, vj). The homogeneous transformation matrix T = [n o a S (ui, vj)] of the plasma torch is calculated using the mapped points S (ui, vj) and the designated posture [n o a]. Using the inverse kinematics of the serial robot, the joint vector θ of the serial robot can be computed. After that, the AM programs are generated and transferred into the serial robotic controller and carried out by the serial robot of Motoman-UP6. The 2D-NURBS curve (square-like) is considered as AM trajectory planning curve, while its corresponding SCCF mapped from the 2D-NURBS curve as AM trajectory. Findings Simulation and experiments show that the preparation of SCCF (mapped from 2D-NURBS curve) on complex NURBS surface using robotic plasma AM is feasible and effective. Originality/value A SCCF mapped from a 2D-NURBS curve is prepared on a complex NURBS surface using the serial robotic plasma AM for the first time. It provides a theoretical and technical basis for plasma AM to produce SCCFs on complex surfaces. With the increasing demand for surface remanufacturing of complex parts, the serial robotic plasma AM of SCCFs on complex NURBS surfaces has a broad application prospect in aero-engine components, high-speed rail power components, nuclear industry components and complex molds.

Traveling load and rotating load can be applied to the tool on the basis of analyzing the relative position between tool and workpiece in assemble model of finite simulation, a finite element model based on trochoid-motion is built. In order to reduce the simulation time, the transition mesh method was used to optimize the finite element model. This paper considers performance parameter of the workpiece, and studied the variation regularities of milling force on the Aeronautical thin-walled by using ABAQUS. At last the experiment shows that the finite element model was verified to be feasible, and the result is reliable.

No abstract

Recent years have witnessed great progress in Synthetic Aperture Radar (SAR) target detection methods based on deep learning. However, these methods generally assume the training data and test data obey the same distribution, which does not always hold when the radar parameters, imaging algorithm, viewpoints, scenes. etc. change in practice. When such a distribution mismatch occurs, it will cause a significant performance drop. Domain adaptation methods provide an effective way to address this problem by transferring knowledge from the source domain (training data) to the target domain (test data). In this work, we proposed an unsupervised Faster R-CNN SAR target detection framework based on domain adaptation, which can improve SAR target detection performance in the unlabeled target domain by borrowing the knowledge of the labeled source domain. Our approach is composed of three stages, pixel domain adaptation (PDA), multilevel feature domain adaptation (MFDA) and iterative pseudo labeling (IPL). By generating transition domain using generative adversarial networks (GANs), the PDA stage can reduce the appearance differences of SAR images. At the MFDA stage, the detector can not only learn the domain invariant global features and instance-level regional features via multilevel adversarial learning in the common feature space but also re-weight the low-level global features according to their relative importance to the target domain. At the IPL stage, we design an iterative pseudo labeling strategy that can select pseudo-labels on instance-level and image-level to encourage the detector to learn more discriminative features of the target domain directly. We evaluate our method using miniSAR and FARADSAR datasets. The experimental results demonstrate the effectiveness of the proposed unsupervised domain adaptation target detection approach. Index Terms -Synthetic Aperture Radar (SAR), target detection, unsupervised domain adaptation, adversarial learning, iterative pseudo labeling.

No abstract