The advancement of next-generation high-frequency communication systems and stealth detection technologies necessitate the development of efficient,multi-spectrum compatible shielding materials.However,the achievement...The advancement of next-generation high-frequency communication systems and stealth detection technologies necessitate the development of efficient,multi-spectrum compatible shielding materials.However,the achievement of simultaneous high efficiency and low reflectivity across microwave,terahertz,and infrared spectra remains a formidable challenge.Herein,a carbonized MXene/polyimide(C-MXene/PI)aerogel material integrating a spatially coupled hierarchically anisotropic structure with stepwise conductivity gradients was constructed.Electromagnetic waves propagate through the top-down vertical disordered horizontal architecture and progressive conductivity gradient of C-MXene/PI aerogel,undergoing stepwise absorption-dissipation-re-dissipation processes.The C-MXene/PI aerogel exhibits an average electromagnetic interference(EMI)shielding effectiveness of91.0 dB in X-band and a reflection coefficient of 0.40.In the terahertz frequency band,the average EMI shielding performance reaches66.2 dB with a reflection coefficient of 0.33.Furthermore,the heterolayered porous architecture of C-MXene/PI aerogels exhibits low thermal conductivity and reduced infrared emissivity,enabling exceptional infrared stealth capability across the 2-16μm wavelength spectrum.This study provides an feasible strategy for constructing low-reflectivity multi-spectrum compatible shielding materials.展开更多
简单线性模型的时间序列预测质量通常超过Transformer等深度模型;而在具有大量通道的数据集上,深度模型尤其是多层感知器(MLP)的性能反而可超过简单线性模型。针对简单线性模型和MLP在时间序列预测中的误差功率谱差异,提出一种基于MLP...简单线性模型的时间序列预测质量通常超过Transformer等深度模型;而在具有大量通道的数据集上,深度模型尤其是多层感知器(MLP)的性能反而可超过简单线性模型。针对简单线性模型和MLP在时间序列预测中的误差功率谱差异,提出一种基于MLP的高频增强型时间序列预测模型HiFNet(High-Frequency Network)。首先,利用MLP在低频段的拟合能力;其次,通过自适应序列分解(ASD)模块及分组线性层解决MLP高频段易过拟合以及通道独立策略不能有效应对通道冗余的问题,从而增强MLP在高频段的鲁棒性;最后,对HiFNet在气象、电力和交通等领域的标准数据集上进行实验。结果表明:HiFNet的均方误差(MSE)在最佳情况下相较于NLinear、RLinear、SegRNN(Segment Recurrent Neural Network)和PatchTST(Patch Time Series Transformer)分别降低了23.6%、10.0%、35.1%和6.5%,而分组线性层通过学习通道相关性的低秩表达减轻了通道冗余的影响。展开更多
基金supported by the Fundamental Research Funds for the Central Universities under No.2024KQ130the National Natural Science Foundation of China(No.52373259)。
文摘The advancement of next-generation high-frequency communication systems and stealth detection technologies necessitate the development of efficient,multi-spectrum compatible shielding materials.However,the achievement of simultaneous high efficiency and low reflectivity across microwave,terahertz,and infrared spectra remains a formidable challenge.Herein,a carbonized MXene/polyimide(C-MXene/PI)aerogel material integrating a spatially coupled hierarchically anisotropic structure with stepwise conductivity gradients was constructed.Electromagnetic waves propagate through the top-down vertical disordered horizontal architecture and progressive conductivity gradient of C-MXene/PI aerogel,undergoing stepwise absorption-dissipation-re-dissipation processes.The C-MXene/PI aerogel exhibits an average electromagnetic interference(EMI)shielding effectiveness of91.0 dB in X-band and a reflection coefficient of 0.40.In the terahertz frequency band,the average EMI shielding performance reaches66.2 dB with a reflection coefficient of 0.33.Furthermore,the heterolayered porous architecture of C-MXene/PI aerogels exhibits low thermal conductivity and reduced infrared emissivity,enabling exceptional infrared stealth capability across the 2-16μm wavelength spectrum.This study provides an feasible strategy for constructing low-reflectivity multi-spectrum compatible shielding materials.
文摘简单线性模型的时间序列预测质量通常超过Transformer等深度模型;而在具有大量通道的数据集上,深度模型尤其是多层感知器(MLP)的性能反而可超过简单线性模型。针对简单线性模型和MLP在时间序列预测中的误差功率谱差异,提出一种基于MLP的高频增强型时间序列预测模型HiFNet(High-Frequency Network)。首先,利用MLP在低频段的拟合能力;其次,通过自适应序列分解(ASD)模块及分组线性层解决MLP高频段易过拟合以及通道独立策略不能有效应对通道冗余的问题,从而增强MLP在高频段的鲁棒性;最后,对HiFNet在气象、电力和交通等领域的标准数据集上进行实验。结果表明:HiFNet的均方误差(MSE)在最佳情况下相较于NLinear、RLinear、SegRNN(Segment Recurrent Neural Network)和PatchTST(Patch Time Series Transformer)分别降低了23.6%、10.0%、35.1%和6.5%,而分组线性层通过学习通道相关性的低秩表达减轻了通道冗余的影响。
