A data mining approach in real-time measurement for polymer additive manufacturing process with exposure controlled projection lithography. (April 2017)

• Record Type: Journal Article
• Authors: Zhao, Xiayun; Rosen, David W.
• Is Part Of: Journal of manufacturing systems. Volume 43:Part 2(2017)
• Page Start: 271
• Page End: 286
• Publication Date: 2017-04
• Subjects: Data mining -- Additive manufacturing -- Real-time process measurement -- Adaptive estimation -- Curve fitting -- Statistical learning -- Robust regression -- Photopolymerization -- Interferometry -- Sensor model; Manufacturing processes -- Periodicals; Production engineering -- Data processing -- Periodicals; Robots, Industrial -- Periodicals; Production, Technique de la -- Informatique -- Périodiques; Robots industriels -- Périodiques; Electronic journals; 670.42
• DOI: 10.1016/j.jmsy.2017.01.005
• Languages: English
• ISSNs: 0278-6125

Abstract

Highlights: A data mining approach for evaluating an interferometric curing monitoring and measuring (ICM&M) sensor model is developed, to enable a real-time measurement method of a photopolymer additive manufacturing process. The ICM&M algorithms are designed and verified to be intelligent, accurate, robust and efficient for handling large volume of stream data with process dynamics and noises. Algorithms parameters effects are studied, and empirical values obtained from experimental observations are incorporated to guarantee realistic solutions derived. The measurement characteristics of ICM&M accuracy, precision, capability and uncertainty are revealed by experiment data analysis. The developed ICM&M method visualizes the process dynamics useful for modeling and control of photopolymerization based additive manufacturing processes. Abstract: Real-time inspection and part dimensions determination during the manufacturing process can improve production of qualified parts. Exposure Controlled Projection Lithography (ECPL) is a bottom-up mask-projection additive manufacturing (AM) process, in which micro parts are fabricated from photopolymers on a stationary transparent substrate. An in-situ interferometric curing monitoring and measuring (ICM&M) system has been developed to infer the output of cured height. Successful ICM&M practice of data acquisition and analysis for retrieving useful information is central to the success of real-time measurement and control for the ECPL process. As the photopolymerization phenomena occur continuously over a range of space and time scales, the ICM&M data analysis is complicated with computation speed and cost. The large amount of video data, which is usually noisy and cumbersome, requires efficient data analysis methods to unleash the ICM&M capability. In this paper, we designed a pragmatic approach of ICM&M data mining to intelligently decipher part height across the cured part. As a data-driven measurement method, the ICM&M algorithms are strengthened by incorporating empirical values obtained from experimental observations to guarantee realistic solutions, and they are particularly useful in real time when limited resource is accessible for online computation. Experimental results indicate that the data-enabled ICM&M method could estimate the height profile of cured parts with accuracy and precision. The study exemplifies that data mining techniques can help realize the desired real time measurement for AM processes, and help unveil more insights about the process dynamics for advanced modeling and control.