Technical Library | 2020-10-08 00:55:22.0
This article presents the development of a stretchable sensor network with high signal-to-noise ratio and measurement accuracy for real-time distributed sensing and remote monitoring. The described sensor network was designed as an island-and-serpentine type network comprising a grid of sensor "islands" connected by interconnecting "serpentines." A novel high-yield manufacturing process was developed to fabricate networks on recyclable 4-inch wafers at a low cost. The resulting stretched sensor network has 17 distributed and functionalized sensing nodes with low tolerance and high resolution. The sensor network includes Piezoelectric (PZT), Strain Gauge(SG), and Resistive Temperature Detector (RTD) sensors. The design and development of a flexible frame with signal conditioning, data acquisition, and wireless data transmission electronics for the stretchable sensor network are also presented. The primary purpose of the frame subsystem is to convert sensor signals into meaningful data, which are displayed in real-time for an end-user to view and analyze. The challenges and demonstrated successes in developing this new system are demonstrated, including (a) developing separate signal conditioning circuitry and components for all three sensor types (b) enabling simultaneous sampling for PZT sensors for impact detection and (c)configuration of firmware/software for correct system operation. The network was expanded with an in-house developed automated stretch machine to expand it to cover the desired area. The released and stretched network was laminated into an aerospace composite wing with edge-mount electronics for signal conditioning, processing, power, and wireless communication.
Technical Library | 2024-04-29 21:19:42.0
Over the years, computer vision and machine learning disciplines have considerably advanced the field of automated visual inspection for Printed Circuit Board (PCB-AVI) assurance. However, in practice, the capabilities and limitations of these advancements remain unknown because there are few publicly accessible datasets for PCB visual inspection and even fewer that contain images that simulate realistic application scenarios. To address this need, we propose a publicly available dataset, "FICS-PCB"1, to facilitate the development of robust methods for PCB-AVI. The proposed dataset includes challenging cases from three variable aspects: illumination, image scale, and image sensor. This dataset consists of 9,912 images of 31 PCB samples and contains 77,347 annotated components. This paper reviews the existing datasets and methodologies used for PCBAVI, discusses challenges, describes the proposed dataset, and presents baseline performances using feature engineering and deep learning methods for PCB component classification.
| 1 |
