熔融沉积成型3D打印的声发射监控理论与方法研究

A closed-loop monitoring method based on sensors is needed for 3D printing in order to ensure the reliability of the printing process and the product quality. This research project is to study the theory and method of monitoring fused deposition modeling (FDM), where acoustic emission (AE) is adopted as the sensing technique. There are 4 main study aspects, which are: 1) To systematically analyze the factors influencing the quality of the FDM 3D printing, put forward an effective monitoring strategy, develop a monitoring and testing system platform, and study the mechanism of AE signals emitted from the 3D printing process for building the theoretical basis of the monitoring by acoustic emission; 2) Study the algorithm of efficient AE signal processing and feature extracting that meet the monitoring and real-time demands, as AE has the characteristics of high sensitivity and big data stream. 3) Aiming at the typical FDM process failures including extruder blockage, material filament breakage, material run out, and etc., study the corresponding monitoring theory and method with the combination of conducting experiments and certain machine learning algorithms; 4) Based on the non-supervised clustering algorithms of data mining, study the method of identifying the typical material defects of 3D printed parts such as material curling, peeling, shrinking and cracking, and predicting the defects’ growing tendency. The results of this project can provide the significant theoretical foundation and technical support for achieving the intelligent monitoring of FDM and improving the product quality.

为了保证3D打印的过程稳定性和产品质量，需要对打印过程实施基于传感器的闭环监控。本项目基于声发射传感器技术，研究熔融沉积成型3D打印的监控理论与方法，主要包括：1）系统地分析熔融沉积成型3D打印的质量影响因素，研究其实时监控策略，构建声发射监控测试平台，探索增材过程中声发射的成因机理，夯实声发射监控的理论基础；2）针对AE信号的高敏感性和大量数据问题，研究可满足监控需求及实时性要求的声发射信号分析与特征提取算法；3）针对打印喷头阻塞、打印材料断丝和材料耗尽等典型的3D打印机过程故障，结合实验和机器学习相关算法研究其实时智能识别方法；4）采用无监督实时聚类算法等数据挖掘工具，研究3D打印产品中的卷曲、翘起、收缩和裂痕等典型缺陷的识别方法，并分析这些缺陷的劣变趋势。本项目的研究成果，可为实现融沉积成型3D打印的智能化监控和提高其打印产品质量提供重要理论依据与技术支撑。

本项目以熔融沉积成型3D打印为对象，以声发射、特征提取和智能识别等为主要技术手段，开展针对其打印过程的质量监控理论与故障诊断方法研究。本项目针对声发射信号数据量大、信息丰富等特点，根据具体的信号特点，分别采用原始波形信号分析和基于波击的参数化方法，提取打印过程中特定工况下的典型特征，建立传感信号与具体工况之间的映射关系，探究传感信号的产生机理；研究传感信号与打印机故障之间的关系，建立一套基于模式识别的、问题驱动的打印机质量监控模型；研究相关的特征提取与模式识别方法，提出一套稳健性和实时性更强的质量监控系统；采用聚类分析等方法，研究传感信号与各类打印产品缺陷之间的关系，分析各类缺陷的劣化趋势；研究典型翘曲变形缺陷下的传感信号参数化特征，提取缺陷发生时信号相关频率成分；结合红外阵列传感器，对打印过程中非动态发展的表面缺陷进行监测，提出一套基于智能识别的缺陷识别方法。本项目的研究成果将为企业生产带有质量监控系统的高端熔融沉积成型打印机提供一套行之有效的方法。.本课题重点分析了我国相关增材制造企业在打印产品质量方面存在的问题，提出建立质量监控系统的需求。在对熔融沉积成型打印过程中可能出现的打印机故障和打印产品缺陷类型进行研究的基础上，构建了基于特定传感信号特征提取和智能模式识别的模型。通过对熔融沉积成型打印过程中基于传感器的在线监测模型的构建，为及时采取修复应对措施、实现打印过程闭环控制奠定了基础；在对打印过程监控研究成果进行分析的基础上，提出基于多传感器的打印喷头和打印工件同步监测的方法，使监控系统更加符合企业的实际情况。