Public Abstract

DE-SC0025550: UMNS: Ultrasonic MEMS Neuromorphic Sensor for Additive Manufacturing In-Process Monitoring

Award Status: Active

Institution: The Board of Regents, University of Nebraska for the University of Nebraska-Lincoln, Lincoln, NE
UEI: HTQ6K6NJFHA6
DUNS: 555456995

Most Recent Award Date: 09/05/2025
Number of Support Periods: 2
PM: Kidd, Timothy

Current Budget Period: 09/01/2025 - 08/31/2026
Current Project Period: 09/01/2024 - 08/31/2028
PI: Alsaleem, Fadi

Supplement Budget Period: N/A

Public Abstract

Additive Manufacturing (AM) has recently gained much attention from the industrial and research community. It has many advantages compared to traditional subtractive methods, such as reducing waste and lead time and producing more complex geometries and shorter product lifecycles. Despite AM's potential, its quality control and high defect probability due to the complex processing parameters and the feedstock quality remain the principal challenge. An added complexity is the requirement to detect these defects using sensor data during manufacturing, not after completion, since the layer-by-layer manufacturing approach characterizes AM. To give a perspective of this complexity, the amount of data these sensors produce during manufacturing competes with Netflix (e.g., in the optical method, 75.1 GB of images per second is produced by a high-speed imaging system) and necessitates ample memory storage and high-speed computing power to store and process the data. To overcome this challenge, this project aims to develop a novel analog computing unit inspired by the efficient analog computing performed by our brain. This computing hardware will be designed to acquire and process acoustic data in real time to detect defects and characterize the quality of AM products during manufacturing. Thus, this new in-situ monitoring of AM processes can detect defects during manufacturing to improve the quality of AM products. This approach is also inherently secure since no data is stored, making it invulnerable to cyberattacks. The University of Nebraska-Lincoln (UNL) team is highly qualified to perform the proposed work. It brings unique skills and expertise in analog computing, acoustic sensing, and AM. Dr. Alsaleem, the PI, has pioneered the idea of analog computing. Co-PI Dr. Turner is an expert on AM, and Co-PI Dr Zhu has expertise in acoustic sensing. The team is joined by Dr. David Mascarenas, an R&D Engineer at Los Alamos National Laboratory (LANL). The proposed work is significant as it presents the first development of an analog-based computer for real-time in-process monitoring of AM products. It will also entail a close collaboration with scientists at the Los Alamos National Lab (LANL) lab working on the critical field of AM. This work also aligns with the DOE's priorities in advanced manufacturing (additive manufacturing) and smart manufacturing. The proposed work would also be relevant to DOE/NNSA applications, including low-power sensors for persistent surveillance, scientific computing at the edge, global security, and low-power tamper evident seals for material accounting and treaty verification. Finally, multiple graduate students will be trained throughout the project in diverse skills such as finite element modeling, microfabrication, and machine learning. At least two former students, supervised by the Co-PI Zhu, joined DOE national labs after graduation.

Additive Manufacturing (AM) has recently gained much attention from the industrial and research community. It has many advantages compared to traditional subtractive methods, such as reducing waste and lead time and producing more complex geometries and shorter product lifecycles. Despite AM's potential, its quality control and high defect probability due to the complex processing parameters and the feedstock quality remain the principal challenge. An added complexity is the requirement to detect these defects using sensor data during manufacturing, not after completion, since the layer-by-layer manufacturing approach characterizes AM. To give a perspective of this complexity, the amount of data these sensors produce during manufacturing competes with Netflix (e.g., in the optical method, 75.1 GB of images per second is produced by a high-speed imaging system) and necessitates ample memory storage and high-speed computing power to store and process the data.

To overcome this challenge, this project aims to develop a novel analog computing unit inspired by the efficient analog computing performed by our brain. This computing hardware will be designed to acquire and process acoustic data in real time to detect defects and characterize the quality of AM products during manufacturing. Thus, this new in-situ monitoring of AM processes can detect defects during manufacturing to improve the quality of AM products. This approach is also inherently secure since no data is stored, making it invulnerable to cyberattacks.

The University of Nebraska-Lincoln (UNL) team is highly qualified to perform the proposed work. It brings unique skills and expertise in analog computing, acoustic sensing, and AM. Dr. Alsaleem, the PI, has pioneered the idea of analog computing. Co-PI Dr. Turner is an expert on AM, and Co-PI Dr Zhu has expertise in acoustic sensing. The team is joined by Dr. David Mascarenas, an R&D Engineer at Los Alamos National Laboratory (LANL). The proposed work is significant as it presents the first development of an analog-based computer for real-time in-process monitoring of AM products. It will also entail a close collaboration with scientists at the Los Alamos National Lab (LANL) lab working on the critical field of AM. This work also aligns with the DOE's priorities in advanced manufacturing (additive manufacturing) and smart manufacturing. The proposed work would also be relevant to DOE/NNSA applications, including low-power sensors for persistent surveillance, scientific computing at the edge, global security, and low-power tamper evident seals for material accounting and treaty verification. Finally, multiple graduate students will be trained throughout the project in diverse skills such as finite element modeling, microfabrication, and machine learning. At least two former students, supervised by the Co-PI Zhu, joined DOE national labs after graduation.