In response to the challenges of small-size defects and low recognition rates in Printed Circuit Boards (PCBs), as well as the need for lightweight detection models that can be embedded in portable devices, this paper proposes an improved defect detection method based on a lightweight shared convolutional head using YOLOv8n. Firstly, the Spatial and Channel reconstruction Convolution (SCConv) is embedded into the Cross Stage Partial with Convolutional Layer Fusion (C2f) structure of the backbone network, which reduces redundant computations and enhances the model’s learning capacity. Secondly, an adaptive feature selection module is integrated to improve the network’s ability to recognize small targets. Subsequently, a Shared Lightweight Convolutional Detection (SLCD) Head replaces the original Decoupled Head, reducing the model’s computational complexity while increasing detection accuracy. Finally, the Weighted Intersection over Union (WIoU) loss function is introduced to provide more precise evaluation results and improve generalization capability. Comparative experiments conducted on a public PCB dataset demonstrate that the improved algorithm achieves a mean Average Precision (mAP) of 98.6% and an accuracy of 99.8%, representing improvements of 3.8% and 3.1%, respectively, over the original model. The model size is 4.1 M, and its FPS is 144.1, meeting the requirements for real-time and lightweight portable deployment.

Establishing a novel design and accurate analytical models for XY nanopositioning stages based on voice coil motor (VCM) actuators is critical to achieving an optimal working performance. To overcome the existing design challenges of 2-degree-of-freedom guiding mechanisms, a four-layer structure composed of L-shaped spatial double parallelogram flexure mechanisms was proposed for the magnetic stage, which exhibits light weight and inhibits parasitic and decoupled motions. The guiding mechanisms were modeled by the compliance matrix method. Thereafter, by combining an electromagnetic model for the VCMs with the equivalent magnetic network method, an electromagnetic–mechanical coupling optimization method with multiple constraints was proposed for the stage to achieve a millimeter-range motion with a maximized natural frequency. The mechanical and electromagnetic performances were then verified by finite element analysis software. The optimized prototype was tested with a stroke of ±3.41 and ±3.08 mm for X axis and Y axis, respectively, a closed-loop resolution of 100 nm for X axis and 150 nm for Y axis, and a resonant frequency of 11.75 Hz for both axes. The tracking of a 0.1 Hz spiral of Archimedes achieved a maximum tracking error of 2.9%.

MEMS actuators rely on the deformation of silicon structures. Using dimensions smaller than dozens of micrometers reveals that the micro-electro-mechanical systems (MEMS) actuators are affected by fabrication inaccuracies, leading to hardly predictable forces and/or actuation results. In this paper, MEMS bistable buckled beam actuators are presented. A series of structures based on pre-shaped buckled beams of lengths ranging from 2 to 4 mm, constant width of 5 μm and actuation stroke ranging from 20 to 100 μm was fabricated. Experimental data show a significant difference with predictions from a conventional analytical model. The model commonly used for buckled beams design assumes a rectangular beam section, but it is not the case of the fabricated beams. Furthermore, only symmetric buckling modes (mode 1, mode 3…) are supposed to exist during snap-through. In this paper, new analytical models have been developed on the basis of the models of the literature to consider the effective beam shape. The first improved analytical model enabled prediction of the MEMS buckled beams mechanical behavior in a 30% margin on the whole range of operation. A second model has been introduced to consider both the effective shape of the beam and centro-symmetric buckling modes. This refined model exhibits the partial suppression of buckling mode 2 by a central shuttle. Therefore, mode 2 and mode 3 coexist at the beginning and the end of snap-through, while mode 3 quickly vanishes due to increasing rotation of the central shuttle to leave exclusive presence of mode 2 near the mid-stroke. With this refined model, the effective force-displacement curve can be predicted in a margin reduced to a few percentages in the center zone of the response curve, allowing the accurate prediction of the position switch force. In addition, the proposed model allows accurate results to be reached with very small calculation time.

This paper investigates the optimization of the inner thermal layout in nanofluid-filled horizontal annular pipes under natural convection conditions. Two-dimensional models of annular pipes are established using a numerical simulation approach. Then, using this as the simulation tool, the layouts of single-, triple-, and quadruple-heat-source annular pipes are optimized. The coordinates of the inner cylinders are defined as the decision variables, and the average Nusselt number (Nu) on the cold wall surface as the objective function. For the single-cylinder model, both the Bayesian optimization algorithm (BOA) and the genetic algorithm (GA) derive the same results: an axisymmetric layout where the single heat source is positioned slightly above the axis of the annulus. However, the BOA takes much less computational time than the GA and, consequently, is chosen for cases with more complex geometry. The optimization layout of the three-cylinder model also shows an axisymmetric distribution, while the result for the four-cylinder model presents a centrally symmetric distribution. Compared to the original average Nu, the optimized ones are enhanced by 17.83%, 8.36%, and 6.18% for single-, triple-, and quadruple-heat-source annular pipes, respectively. The results of this study can be used for guiding the layout design and optimization of the nanofluid-filled exchangers with multi-inner heat sources.