1 December 2023 in IEEE Sensors Journal
This paper reports a dual-mode switching of highly sensitive thermal flow sensors based on calorimetric mode that is more suitable for microflow measurements. Upstream heating and downstream heating modes (which are both calorimetric) are adopted in this paper. The sensor consists of two vertically stacked suspended membrane MEMS chips. A gas micro channel is fabricated inside the chip and the chip is mounted perpendicular to the flow direction. Due to the Laval effect, the gas flowing through the internal channels of the chip increases the gas flow velocity and improves the heat exchange efficiency between the chip and the gas. Moreover, the dual-chip stacked design increases the thermal resistance of solid heat conduction and the heat exchange area of the gas, thus improving its sensitivity. Simulation results show that the gas flow velocity and heat exchange efficiency increased seven times and three times, respectively. Experimental results show that the sensor achieves high sensitivity measurements of 0.77 mV/(mL/min) for 0-18 mL/min and 0.16 mV/(mL/min) for 18-100 mL/min over the entire range by dual-mode switching. The response time is 25 ms. To the best of the authors’ knowledge, this is the first report of a thermal flow sensor with dual-mode switching of upstream and downstream heating in calorimetric mode. In addition, the sensor can be widely used in semiconductors, hydrogen energy, scientific instruments and other fields.
1 May 2023 in IEEE Transactions on Cybernetics
Most existing approaches of attributed network embedding often combine topology and attribute information based on the homophily assumption. In many real-world networks, such an assumption does not hold since the nodes are usually associated with many noisy or irrelevant attributes. To tackle this issue, we propose a noise-resistant graph embedding method, called NGE, by leveraging the subspace clustering information (i.e., the formation of communities is driven by different latent features in distinct subspaces). Specifically, we first construct a tensor to represent a given attributed network and then map it into different feature subspaces to capture community structure via tensor decomposition. For structure embedding, the link-level and community-level constraints are imposed. For attribute embedding, the feature-selection constraint is used to reinforce the relationship between topology and noise-removal attributes. By learning structure and attribute embedding with subspace clustering information, NGE can benefit both community detection, link prediction, and node classification. Extensive experimental results have demonstrated the superiority of NGE over many state-of-the-art approaches.