面向微束等离子选区熔化的在线检测系统

为解决微束等离子选区熔化过程中的自动化检测问题,开发出一套面向微束等离子选区熔化的在线检测系统。该系统以每一层的成形金属为研究对象,基于深度学习目标检测原理,对金属成形表面进行缺陷检测,并根据检测结果控制设备的启停,确保每一层的成形质量达到要求。同时,通过双目相机和线激光的配合,利用线阵扫描的方式对成形表面形貌进行三维重建,并显示在自主开发的可视化软件上,实现了缺陷检测和三维重建相结合的检测方案。实验结果表明：该系统能够快速、精确的检测出成形过程中产生的缺陷,实时性强、漏检率低,并将结果及时反馈到系统,实现了成形过程的主动、有效调控。

On-line detection system for microbeam plasma selective melting

In order to solve the problem of shape control in the process of micro-beam plasma selective melting, a set of on-line detection system for microbeam selective melting was developed. The system takes each layer of formed metal as the research object, based on the principle of deep learning target detection, detects defects on the metal forming surface image, and controls the start and stop of the equipment according to the detection results to ensure that the forming quality of each layer meets the requirements. At the same time, through the cooperation of the binocular camera and the line laser, the three-dimensional reconstruction of the formed surface topography is carried out by means of linear array scanning, and displayed on the self-developed visualization software, realizing the detection scheme combining defect detection and three-dimensional reconstruction. The experimental results show that the system can quickly and accurately detect the defects generated in the forming process, with strong real-time performance and low missed detection rate, and the results are fed back to the system in time, realizing the active and effective control of the shape.