电子束选区熔化技术具有成形能量密度高、成形热应力小等优点，在航空航天、生物医疗、汽车等领域具有重要应用价值。电子束选区熔化涉及复杂的电子束-粉末作用，对工艺参数和工况十分敏感，易产生缺陷和吹粉异常，目前其稳定性和可重复性仍有待提高，亟需原位传感技术来监测成形质量和工艺稳定性。但电子束选区熔化过程的蒸镀污染、高温辐射和射线辐射限制了传统基于视觉成像的监测方式的使用。基于电子探测的电子成像技术有效地克服了上述恶劣工况，但现有的电子成像系统无法表达丰富的成形层形貌。本文根据电子束-材料作用过程的辐射电子发射机理，提出了近场、侧向、对称布置的双探测板电子成像方案，实现了对电子束选区熔化成形层形貌的有效监测。所搭建系统探测的电子信号与表面高度梯度存在显著线性关系，能够反映表面的形貌变化。此外，该系统能够有效捕捉目标表面的孔隙特征，孔隙特征分辨率优于100 μm，且具有显著的材料对比度。双探测板电子成像系统探测不同方位的辐射电子信号，构造了具有不同光影效果的电子图像，为成形层形貌特征识别提供了便利。本文提出了基于双通道信号融合的特征识别方法，实现了对成形层孔隙缺陷和起伏形貌的分离。并根据融合信号实现了对成形截面的提取和对截面稳定性的量化表征，截面稳定性指标可以有效地揭示工艺稳定性；此外，提出了基于融合信号的缺陷特征识别和孔隙率表征方法，缺陷分析结果与CT检测结果基本一致。电子束选区熔化工艺目前缺少原位形貌测量方法，电子探测技术提供了一种有效的解决方案。本文提出了基于双通道电子信号的原位形貌重构方法，实现了对成形层形貌的原位检测，检测误差达到±50μm（>90%）。此外，本文根据重构高度分布基于形貌特征分解方法实现了成形表面质量表征和形貌特征识别，为执行缺陷修复时的路径规划提供了依据。吹粉是电子束选区熔化工艺的典型失效形式，但目前缺少有效的监测手段。本文提出了基于电子探测的吹粉监测方法，研究了吹粉过程的电子信号特征，验证了根据初期电子信号衰减特征进行快速吹粉检测的可行性，为设计实时吹粉反馈控制系统提供了依据。此外，本文还利用高速摄影技术和数值仿真对吹粉机理进行了研究，促进了对电子束-粉末作用物理过程的认识。

Electron beam selective melting (EBSM) shows a high application value in some vital industrial sectors like aerospace, biomedical and automatic due to its advantages such as high energy density and low thermal stress during the fabrication. EBSM, involving complex electron beam-powder interaction, is very sensitive to the process parameters and working conditions, thus prone to process defects and smoking erroneous. Thus, its process stability and repeatability still need to be improved. In-situ sensing techniques are urgently required to monitor the fabrication quality and process stability. However, the metal evaporation, as well as the high temperature radiation and X-radiation during process limits the application of traditional optical-based monitoring methods.The electronic imaging technique based on electron detection effectively overcomes the above-mentioned complex working conditions. However, the existing electronic imaging system cannot describe the plentiful morphology of the deposited layers. A near-field, lateral, and symmetrical layout of electronic imaging system with two detection plates was proposed according to the emission mechanism of the feedback electrons. The effective morphology monitoring of the deposited layers was realized. The developed system presents a significant correlation between the electronic signal and surface gradient, thereby showing a strong sensitivity to surface topography. Besides, the electronic imaging system can effectively capture the pore defects with a resolution better than 100 μm, and also shows a high material contrast.The electronic imaging system with the dual detections collects the feedback electron in different directions, thus constructing electronic images with different light and shadow effects, and providing convenience for the recognition of the various features in the deposited layers. A feature separation method based on dual-channel signal fusion was proposed. The separation of pore defects and undulations in the deposited layers was achieved. A method of segmenting the deposited section according to the fused signal was proposed, benefiting for the quantitative characterization of the layerwise fabrication stability. The used section stability index can effectively reveal the process stability. In addition, another method was proposed to identify the pore defects and analyze the porosity based on the fused signal. The result of pore identification matches the result of the computed tomography inspection. EBSM process lacks in-situ surface measurement methods. The electronic detection technique provides an effective solution. A surface reconstruction method based on the dual-channel electronic signals was proposed. The in-situ surface measurement of the deposited surface was realized with an accuracy of ±50μm (>90%). Additionally, based on the reconstructed surface topography, a method for characterizing the surface quality of the fabricated layers and identifying various features was proposed, providing a basis for the path planning in online defect repair.Smoking is a typical failure form of the EBSM process, lacking effective monitoring methods. Therefore, a smoking detection method based on the electronic detection was proposed. The electronic signal pattern of the powder smoking was analyzed, showing a feasibility of fast smoking detection according to the signal attenuation at the early stage of smoking process. The smoking pattern in electronic signal provides a basis for the design of a real-time smoking detection system. In addition, this paper also adopted the high-speed imaging and numerical simulation means to study the mechanism of powder smoking, achieving a promote for the understanding of the physical process of electron beam-powder interaction.