Methods to calculate fluid density and viscosity using a micro-cantilever and based on the resonance principle were put forward. Their measuring mechanisms were analyzed and the theoretical equations to calculate the density and viscosity were deduced. The fluid-solid coupling simulations were completed for the micro-cantilevers with different shapes. The sensing chips with micro-cantilevers were designed based on the simulation results and fabricated using the micro electromechanical systems (MEMS) technology. Finally, the MEMS resonant sensor was packaged with the sensing chip to measure the densities and viscosities of eight different fluids under the flexural and torsional vibrating modes separately. The relative errors of the measured densities from 600 kg/m3 to 900 kg/m3 and viscosities from 200 μPa·s to 1000 μPa·s were calculated and analyzed with different microcantilevers under various vibrating modes. The experimental results showed that the effects of the shape and vibrating mode of micro-cantilever on the measurement accuracies of fluid density and viscosity were analyzed in detail.
Molecular dynamics (MD) simulations were applied to study the fundamental mechanism of nanoscale grinding with a modeled tool trajectory of straight lines. Nevertheless, these models ignore curvature changes of actual tool paths, which need optimization to facilitate understanding of the underlying science of the machining processes. In this work, a three-dimensional MD model considering the effect of tool paths was employed to investigate distributions of stresses including hydrostatic stress, von Mises stress, normal and shear stresses during atomic grinding. Simulation results showed that average values of the stresses are greatly influenced by the radius of the tool trajectory and the grinding depth. Besides the averaged stresses, plane stress distribution was also analyzed, which was obtained by intercepting stresses on the internal planes of the workpiece. For the case of a grinding depth of 25 Å and an arc radius 40 Å, snapshots of the stresses on the X–Y, X–Z and Y–Z planes showed internal stress concentration. The results show that phase transformation occurred from α- silicon to β- silicon in the region with hydrostatic stress over 8 GPa. Moreover, lateral snapshots of the three-dimensional stress distribution are comprehensively discussed. It can be deduced from MD simulations of stress distribution in monocrystalline silicon with the designed new model that a curved tool trajectory leads to asymmetric distribution and concentration of stress during atomic-scale grinding. The analysis of stress distribution with varying curve geometries and cutting depths can aid fundamental mechanism development in nanomanufacturing and provide theoretical support for ultraprecision grinding.
A novel method, which was called a slope method, has been proposed to measure fluid density by the micro-cantilever sensing chip. The theoretical formulas of the slope method were discussed and established when the micro-cantilever sensing chip was under flexural and torsional vibrations. The slope was calculated based on the fitted curve between the excitation and output voltages of sensing chip under the nonresonant status. This measuring method need not sweep frequency to find the accurate resonant frequency. Therefore, the fluid density was measured easily based on the calculated slope. In addition, the micro-cantilver was drived by double sided excitation and free end excitation to oscillate under flexural and torsional vibrations, respectively. The corresponding experiments were carried out to measure the fluid density by the slope method. The measurement results were also analyzed when the sensing chip was under flexural and torsional nonresonant vibrations separately. The measurement accuracies under these vibrations were all better than 1.5%, and the density measuring sensitivity under torsional nonresonant vibration was about two times higher than that under flexural nonresonant vibration.
27 February 2025 in IEEE Internet of Things Journal
For deep neural network (DNN) deployment on Internet of Things (IoT) devices, pipeline-based inference coordinates diverse computing resources in heterogeneous multiprocessor system-on-chips (HMPSoCs) to achieve efficient execution. However, determining the optimal pipeline configuration is challenging due to the exponentially expanding search space and intricate layer-wise dependencies. Existing methods formulate this as a single-shot optimization problem, which struggles to efficiently explore the search space and incurs substantial resource overhead from repeated evaluations of suboptimal configurations. This article proposes PipeMCTS, which reformulates pipeline deployment as a sequential optimization problem solved via Monte Carlo tree search (MCTS). By incrementally constructing the search tree in a layer-wise manner, PipeMCTS effectively prunes unpromising branches and accelerates the search process. PipeMCTS incorporates two key components to enhance search efficiency: 1) a temperature-controlled selection strategy that balances exploration and exploitation in the complex search space and 2) an uncertainty-aware simulation strategy that accelerates search convergence through Gaussian process (GP)-guided evaluation. Experimental results demonstrate that PipeMCTS achieves a 66.42% improvement in throughput compared to arm compute library (ARM-CL).
1 January 2019 in Intelligent Automation and Soft Computing
Subsurface damage of fused silica optics is one of the major factors restricting the performance of optical systems. The densification-affected deformation and fracture in fused silica under a sliding contact are investigated in this study, via three-dimensional finite element analysis (FEA). The finite element models of scratching with 70.3° conical and Berkovich indenters are established. A refined elliptical constitutive model is used to consider the influence of densification. The finite element models are experimentally verified by elastic recovery, and theoretically verified by hardness ratio. Results of densification and plastic deformation distributions indicate that the accuracy of existent sliding stress field models may be improved if the spherical/cylindrical yield region is replaced by an ellipsoid/cylindroid, and the embedding of the yield region is considered. The initiation sequence, and the locations and stages of radial, median, and lateral cracks are discussed by analyzing the predicted sliding stress fields. Median and radial cracks along the sliding direction tend to be the first cracks that emerge in the sliding and unloading stages, respectively. They coalesce to form a big median–radial crack that penetrates through the entire yield region. The fracture behavior of fused silica revealed in this study is essential in the low-damage machining of fused silica optics.