期刊文献+
共找到444篇文章
< 1 2 23 >
每页显示 20 50 100
Tailoring a traditional Chinese medicine prescription for complex diseases:A novel multi-targets-directed gradient weighting strategy
1
作者 Zhe Yu Teng Li +11 位作者 Zhi Zheng Xiya Yang Xin Guo Xindi Zhang Haoying Jiang Lin Zhu Bo Yang Yang Wang Jiekun Luo Xueping Yang Tao Tang En Hu 《Journal of Pharmaceutical Analysis》 2025年第4期804-816,共13页
Traditional Chinese medicine(TCM)exerts integrative effects on complex diseases owing to the characteristics of multiple components with multiple targets.However,the syndrome-based system of diagnosis and treatment in... Traditional Chinese medicine(TCM)exerts integrative effects on complex diseases owing to the characteristics of multiple components with multiple targets.However,the syndrome-based system of diagnosis and treatment in TCM can easily lead to bias because of varying medication preferences among physicians,which has been a major challenge in the global acceptance and application of TCM.Therefore,a standardized TCM prescription system needs to be explored to promote its clinical application.In this study,we first developed a gradient weighted disease-target-herbal ingredient-herb network to aid TCM formulation.We tested its efficacy against intracerebral hemorrhage(ICH).First,the top 100 ICH targets in the GeneCards database were screened according to their relevance scores.Then,SymMap and Traditional Chinese Medicine Systems Pharmacology(TCMSP)databases were applied to find out the target-related ingredients and ingredient-containing herbs,respectively.The relevance of the resulting ingredients and herbs to ICH was determined by adding the relevance scores of the corresponding targets.The top five ICH therapeutic herbs were combined to form a tailored TCM prescriptions.The absorbed components in the serum were detected.In a mouse model of ICH,the new prescription exerted multifaceted effects,including improved neurological function,as well as attenuated neuronal damage,cell apoptosis,vascular leakage,and neuroinflammation.These effects matched well with the core pathological changes in ICH.The multi-targets-directed gradient-weighting strategy presents a promising avenue for tailoring precise,multipronged,unbiased,and standardized TCM prescriptions for complex diseases.This study provides a paradigm for advanced achievements-driven modern innovation in TCM concepts. 展开更多
关键词 Traditional Chinese medicine Prescription formulation gradient weighting strategy Multiple targets Intracerebral hemorrhage
暂未订购
Inversion of gravity gradient data based on spatial gradient weighting
2
作者 JIANG DanDan YU Ping +1 位作者 LIN Song GAO Xiuhe 《Global Geology》 2018年第4期245-251,共7页
Compared with traditional gravity measurement data,gravity gradient tensor data contain more high frequency information,which can be used to understand the earth's interior structure,mineral resources distribution... Compared with traditional gravity measurement data,gravity gradient tensor data contain more high frequency information,which can be used to understand the earth's interior structure,mineral resources distribution etc. In this study,the authors present an algorithm for inverting gravity gradiometer data to recover the three-dimensional( 3-D) distributions of density. Spatial gradient weighting was used to constrain the extent of the body horizontally and vertically. A more accurate inversion result can be obtained by combining the prior information into the weighting function and applying it in inversion. This method was tested on synthetic models and the inverted results showed that the resolution was significantly improved. Moreover,the algorithm was applied to the inversion of empirical data from a salt dome located in Texas,USA,which demonstrated the validity of the proposed method. 展开更多
关键词 GRAVITY gradient DATA SPATIAL gradient weighting 3-D INVERSION
在线阅读 下载PDF
GRADIENT ESTIMATES AND ENTROPY FORMULAE FOR WEIGHTED p-HEAT EQUATIONS ON SMOOTH METRIC MEASURE SPACES 被引量:4
3
作者 王宇钊 杨杰 陈文艺 《Acta Mathematica Scientia》 SCIE CSCD 2013年第4期963-974,共12页
Let (M,g, e^-fdv) be a smooth metric measure space. In this paper, we con- sider two nonlinear weighted p-heat equations. Firstly, we derive a Li-Yau type gradient estimates for the positive solutions to the followi... Let (M,g, e^-fdv) be a smooth metric measure space. In this paper, we con- sider two nonlinear weighted p-heat equations. Firstly, we derive a Li-Yau type gradient estimates for the positive solutions to the following nonlinear weighted p-heat equationand f is a smooth function on M under the assumptionthat the m-dimensional nonnegative Bakry-Emery Ricci curvature. Secondly, we show an entropy monotonicity formula with nonnegative m-dimensional Bakry-Emery Ricci curva- ture which is a generalization to the results of Kotschwar and Ni [9], Li [7]. 展开更多
关键词 gradient estimates weighted p-heat equation entropy monotonicity formula m-Bakry-t^mery Ricci curvature
在线阅读 下载PDF
Hamilton-Souplet-Zhang's gradient estimates for two weighted nonlinear parabolic equations 被引量:1
4
作者 MA Bing-qing HUANG Guang-yue 《Applied Mathematics(A Journal of Chinese Universities)》 SCIE CSCD 2017年第3期353-364,共12页
In this paper, we consider gradient estimates for positive solutions to the following weighted nonlinear parabolic equations on a complete smooth metric measure space with only Bakry-Émery Ricci tensor bounded be... In this paper, we consider gradient estimates for positive solutions to the following weighted nonlinear parabolic equations on a complete smooth metric measure space with only Bakry-Émery Ricci tensor bounded below: One is $${u_t} = {\Delta _f}u + au\log u + bu$$ with a, b two real constants, and another is $${u_t} = {\Delta _f}u + \lambda {u^\alpha }$$ with λ, α two real constants. We obtain local Hamilton-Souplet-Zhang type gradient estimates for the above two nonlinear parabolic equations. In particular, our estimates do not depend on any assumption on f. 展开更多
关键词 Hamilton’s gradient estimate Souplet-Zhang’s gradient estimate weighted nonlinear parabolic equation Bakry-Émery Ricci tensor
在线阅读 下载PDF
A sharp gradient estimate for the weighted p-Laplacian 被引量:2
5
作者 WANG Lin-feng ZHU Yue-ping 《Applied Mathematics(A Journal of Chinese Universities)》 SCIE CSCD 2012年第4期462-474,共13页
Let M be an n-dimensional complete noncompact Riemannian manifold with sectional curvature bounded from below, dμ = e^h(x) dV(x) the weighted measure and △μ,p the weighted p-Laplacian. In this paper we consider... Let M be an n-dimensional complete noncompact Riemannian manifold with sectional curvature bounded from below, dμ = e^h(x) dV(x) the weighted measure and △μ,p the weighted p-Laplacian. In this paper we consider the non-linear elliptic equation △μ,pu=-λμ,p|u|^p-2ufor p ∈ (1, 2). We derive a sharp gradient estimate for positive smooth solutions of this equation. As applications, we get a Harnack inequality and a Liouville type theorem.. 展开更多
关键词 weighted p-Laplacian gradient estimate Harnack inequality Liouville theorem.
在线阅读 下载PDF
Multi-Innovation Gradient Iterative Locally Weighted Learning Identification for A Nonlinear Ship Maneuvering System 被引量:3
6
作者 BAI Wei-wei REN Jun-sheng LI Tie-shan 《China Ocean Engineering》 SCIE EI CSCD 2018年第3期288-300,共13页
This paper explores a highly accurate identification modeling approach for the ship maneuvering motion with fullscale trial. A multi-innovation gradient iterative(MIGI) approach is proposed to optimize the distance me... This paper explores a highly accurate identification modeling approach for the ship maneuvering motion with fullscale trial. A multi-innovation gradient iterative(MIGI) approach is proposed to optimize the distance metric of locally weighted learning(LWL), and a novel non-parametric modeling technique is developed for a nonlinear ship maneuvering system. This proposed method’s advantages are as follows: first, it can avoid the unmodeled dynamics and multicollinearity inherent to the conventional parametric model; second, it eliminates the over-learning or underlearning and obtains the optimal distance metric; and third, the MIGI is not sensitive to the initial parameter value and requires less time during the training phase. These advantages result in a highly accurate mathematical modeling technique that can be conveniently implemented in applications. To verify the characteristics of this mathematical model, two examples are used as the model platforms to study the ship maneuvering. 展开更多
关键词 multi-innovation gradient iterative(MIGI) locally weighted learning(LWL) IDENTIFICATION nonlinearship maneuvering full-scale trial
在线阅读 下载PDF
Gradient effect of road transportation on economic development in different geomorphic regions 被引量:5
7
作者 YANG Xue-ting FANG Yi-ping +1 位作者 QIU Xiao-ping ZHU Fu-biao 《Journal of Mountain Science》 SCIE CSCD 2018年第1期181-197,共17页
The economic benefits of transport infrastructure investment have been widely accepted.However,the varying influence of road transport development across vertical space has rarely been discussed.Taking Sichuan provinc... The economic benefits of transport infrastructure investment have been widely accepted.However,the varying influence of road transport development across vertical space has rarely been discussed.Taking Sichuan province in China as case study area where the landform is diverse and complex,administrative counties were categorized into 4 main types:plain counties,hill counties,mountain counties,and plateau counties.Using statistical data during 2006-2014,theperformanceofeconomic development and transport construction level in the four types of counties are discussed.Subsequently,the heterogeneous effect of each grade road on economy was calculated by local regression model(GWR).The results indicate that plain counties largely surpassed the other geomorphic counties in economic development level,while the gradient gap among them was on the decline.Similarly,distribution of transport infrastructure presented a decreasing trend from the low plain counties to high plateau counties.Regional imbalances were mainly reflected in the County road and Village road.Regarding the changes of regional gaps,National&Provincial roads and County roads were constantly expanding,whereas the disparity of Village road was slowly narrowing over time.Particularly noteworthy was the non-stationary economic influence of traffic factors across vertical gradients.On average,National&Provincial roads generated higher benefits in the high elevation regions than the lowlands.In contrast,County road and Village road were found to be more effective in promoting economic development in plains.With regard to local estimates of traffic factors,coefficients in mountain counties exhibited larger fluctuation ranges than other geomorphic units.The conclusions provide a basis for government decisionmaking in a more reasonable construction arrangement of road facilities and sustainable economic development. 展开更多
关键词 Geomorphic counties Road transportation Economic development Geographically weighted regression gradient variation
原文传递
Patch-based vehicle logo detection with patch intensity and weight matrix 被引量:3
8
作者 刘海明 黄樟灿 Ahmed Mahgoub Ahmed Talab 《Journal of Central South University》 SCIE EI CAS CSCD 2015年第12期4679-4686,共8页
A patch-based method for detecting vehicle logos using prior knowledge is proposed.By representing the coarse region of the logo with the weight matrix of patch intensity and position,the proposed method is robust to ... A patch-based method for detecting vehicle logos using prior knowledge is proposed.By representing the coarse region of the logo with the weight matrix of patch intensity and position,the proposed method is robust to bad and complex environmental conditions.The bounding-box of the logo is extracted by a thershloding approach.Experimental results show that 93.58% location accuracy is achieved with 1100 images under various environmental conditions,indicating that the proposed method is effective and suitable for the location of vehicle logo in practical applications. 展开更多
关键词 vehicle logo detection prior knowledge gradient extraction patch intensity weight matrix background removing
在线阅读 下载PDF
The UNIFORM C^(0)ESTIMATE AND WEIGHTED ESTIMATE OF GENERALIZED CHRISTOFFEL-MINKOWSKI PROBLEMS
9
作者 ZHANG Jin-hu 《数学杂志》 2024年第5期397-405,共9页
In this paper,we consider generalized Christo®el-Minkowski problems as followsσ_(k)(u_(ij)+uδ_(ij))/σ_(l)(u_(ij)+uδ_(ij))=|u^(p-1)f(x),x∈S^(n),where 0≤l≤k≤n,p-1>0 and f is positive,and we establish the... In this paper,we consider generalized Christo®el-Minkowski problems as followsσ_(k)(u_(ij)+uδ_(ij))/σ_(l)(u_(ij)+uδ_(ij))=|u^(p-1)f(x),x∈S^(n),where 0≤l≤k≤n,p-1>0 and f is positive,and we establish the weighted gradient estimate and uniform C^(0)estimate for the positive convex even solutions,which is a generalization of Guan-Xia[1]and Guan[2]. 展开更多
关键词 weighted gradient estimate convex solution minkowski type problem
在线阅读 下载PDF
SOLUTIONS TO NONLINEAR ELLIPTIC EQUATIONS WITH A GRADIENT
10
作者 王影 王明新 《Acta Mathematica Scientia》 SCIE CSCD 2015年第5期1023-1036,共14页
In this article, we consider existence and nonexistence of solutions to problem {-△pu+g(x,u)|↓△|^p=f in -Ω,u=0 on Ω with 1〈p〈∞ where f is a positive measurable function which is bounded away from 0 in Ω,... In this article, we consider existence and nonexistence of solutions to problem {-△pu+g(x,u)|↓△|^p=f in -Ω,u=0 on Ω with 1〈p〈∞ where f is a positive measurable function which is bounded away from 0 in Ω, and the domain Ω is a smooth bounded open set in R^N(N≥2). Especially, under the condition that g(x, s) = 1/|s|^α (α〉0) is singular at s = 0, we obtain that α 〈 p is necessary and sufficient for the existence of solutions in W0^1,p(Ω) to problem (0.1) when f is sufficiently regular. 展开更多
关键词 quasilinear elliptic equations existence and nonexistence gradient terms singular weights
在线阅读 下载PDF
Gradient Estimates for a Simple Lichnerowicz Equation on Compact Manifolds
11
作者 邢杰 《Chinese Quarterly Journal of Mathematics》 2015年第3期317-322,共6页
Let(M,g) be a smooth compact Riemannian manifold of dimension n.Denote△f=△-▽f.▽ the weighted Laplacian operator,where f is a smooth real valued function on M.When N is finite and the N-Bakry-Emery Ricci tensor is ... Let(M,g) be a smooth compact Riemannian manifold of dimension n.Denote△f=△-▽f.▽ the weighted Laplacian operator,where f is a smooth real valued function on M.When N is finite and the N-Bakry-Emery Ricci tensor is bounded from below by a constant,we establish local gradient estimates for positive solutions of the following simple Lichnerowicz equation△fu+cu^(-α)=0 on a compact Riemannian manifold,where α is a positive constant and c is a smooth function. 展开更多
关键词 gradient ESTIMATE Lichnerowicz EQUATION weightED LAPLACIAN
在线阅读 下载PDF
基于SVM模型的亚像素位移测量方法的研究 被引量:1
12
作者 沈澍 孙磊磊 +2 位作者 孙义杰 张浩 王森 《小型微型计算机系统》 北大核心 2025年第5期1156-1160,共5页
亚像素位移测量算法常见的有曲面拟合法、梯度法和插值法等,论文介绍了一种基于支持向量机(SVM)模型的亚像素位移测量方法,提出一种用于计算物体移动前后两幅图像的相关值来表征亚像素位移的梯度加权求和法.利用计算机模拟生成一系列以0... 亚像素位移测量算法常见的有曲面拟合法、梯度法和插值法等,论文介绍了一种基于支持向量机(SVM)模型的亚像素位移测量方法,提出一种用于计算物体移动前后两幅图像的相关值来表征亚像素位移的梯度加权求和法.利用计算机模拟生成一系列以0.001pixel步调的图像,其中70%作为训练集,30%作为测试集,为了检验该模型抗噪性,在生成的图像中添加不同方差的高斯噪声,并将其与曲面拟合法和梯度法进行比较,结果表明在精度要求为0.01pixel时,采用SVM法在保证高精度的情况下其抗噪性优于曲面拟合法和梯度法,其噪声方差上限为0.015,具有很好的鲁棒性可用于实际物体位移的高精度测量. 展开更多
关键词 支持向量机(SVM) 亚像素位移 数字图像相关法(DICM) 梯度加权求和法 高斯噪声
在线阅读 下载PDF
井下煤炭运输多环节复杂背景下高精度煤矸识别方法
13
作者 栾亨宣 安乐 +7 位作者 田莹 顾颉颖 张强 陈宏伟 付翔 胡成军 崔志芳 陈锐 《煤炭科学技术》 北大核心 2025年第9期459-468,共10页
煤矸智能分选是发展煤炭智能绿色开采的关键技术之一。准确识别煤矸石是智能分选的先决条件,如何克服井下复杂环境的影响,已成为影响机器视觉识别效果的现实难题。针对井下煤炭运输系统,构建具有单一、二混合和三混合背景的11类图像数据... 煤矸智能分选是发展煤炭智能绿色开采的关键技术之一。准确识别煤矸石是智能分选的先决条件,如何克服井下复杂环境的影响,已成为影响机器视觉识别效果的现实难题。针对井下煤炭运输系统,构建具有单一、二混合和三混合背景的11类图像数据集,包含水渍、煤粉、碎煤与碎矸石及运输设备部件等背景干扰因素。提出了一种融合多种注意力与残差连接的高精度煤矸智能识别方法,高分辨率阶段使用残差卷积块快速生成高质量标记(token),后续阶段使用级联的Channel Spatial Swin Transformer Block(CSSTB)进行深层表征学习。为提升模型对背景噪声的抗干扰能力,网络集成全局、通道、空间多种注意力机制,增强特征表达的鲁棒性。CSSTB中利用基于负斜率特性强化稀疏激活的LeakyReLU线性注意力机制建模全局信息,通过卷积注意力机制模块(CBAM)优化模型注意力分布,提升模型泛化能力。此外,考虑到煤、矸石和输送设备零部件的尺度差异,跨阶段使用残差连接以增强多尺度特征的通信和信息流通。结果表明:所提模型在单一、二混合和三混合背景上的平均准确度达到95.06%、97.77%、95.65%,相较于基线网络Swin Transformer-Tiny分别提高7.01%、4.83%、1.03%。可视化试验表明,对比模型在水渍、暗光和反光等复杂背景干扰下难以准确区分煤和矸石,而所提出模型能够精准聚焦于煤和矸石的关键特征区域,抗干扰能力强。研究结果为井下原煤运输中的煤矸高效分选提供了理论参考。 展开更多
关键词 深度学习 煤矸识别 井下环境 注意力机制 梯度加权类激活热图
在线阅读 下载PDF
基于多传感器融合的液压支架位姿精确感知方法 被引量:2
14
作者 马长青 李旭阳 +3 位作者 李峰 毛俊杰 魏祥宇 马肖杨 《工矿自动化》 北大核心 2025年第4期114-119,共6页
为精确感知扰动环境下液压支架位姿信息,提出了一种基于多传感器融合的液压支架位姿精确感知方法。首先,在液压支架顶梁、掩护梁、后连杆和底座4个构件上部署九轴姿态传感器,利用其陀螺仪、加速度计和磁力计分别解算出其所在构件的横滚... 为精确感知扰动环境下液压支架位姿信息,提出了一种基于多传感器融合的液压支架位姿精确感知方法。首先,在液压支架顶梁、掩护梁、后连杆和底座4个构件上部署九轴姿态传感器,利用其陀螺仪、加速度计和磁力计分别解算出其所在构件的横滚角、俯仰角和偏航角等位姿数据;然后,通过无迹卡尔曼滤波(UKF)算法和梯度下降(IGD)算法(IGD-UKF算法)对位姿数据进行滤波处理,降低扰动因素对位姿数据的干扰;最后,采用自适应加权融合算法对滤波处理后的液压支架顶梁和底座的偏航角和横滚角数据进行融合处理,消除外界振动、噪声等因素引起的液压支架顶梁和底座传感器数据误差。对施加扰动下液压支架顶梁低头和抬头、底座低头和抬头、液压支架左倾和右倾、液压支架左偏和右偏等工况下顶梁、掩护梁、后连杆和底座的位姿进行感知实验,结果表明:经IGD-UKF算法处理后的数据曲线波动趋于平缓,在抑制振荡、减小振幅上的效果明显;液压支架偏航角误差为0.001 8~0.025 1°,平均绝对误差为0.004 8°,横滚角误差为0.001 4~0.028 1°,平均绝对误差为0.004 7°,实现了扰动环境下液压支架位姿的精确感知。 展开更多
关键词 液压支架 支架位姿感知 多传感器融合 无迹卡尔曼滤波 梯度下降 自适应加权融合 九轴姿态传感器
在线阅读 下载PDF
共同富裕水平的评价与预测——基于机器学习算法 被引量:2
15
作者 薛洁 武欣妍 《调研世界》 2025年第4期17-32,共16页
共同富裕是社会主义的本质要求,研究2016—2022年共同富裕在全国31个省(区、市)的发展,对于全面建成社会主义现代化强国有参考作用。本文从经济质量、协调发展、生活丰富、社会稳定、公共服务、美丽家园等六个方面,选择41个评价指标,构... 共同富裕是社会主义的本质要求,研究2016—2022年共同富裕在全国31个省(区、市)的发展,对于全面建成社会主义现代化强国有参考作用。本文从经济质量、协调发展、生活丰富、社会稳定、公共服务、美丽家园等六个方面,选择41个评价指标,构建共同富裕评价指标体系,并利用CRITIC权重法对各省(区、市)发展水平进行测算,采用K-Means++聚类算法对一级指标得分结果进行等级分类。随后创新性地通过梯度提升算法对各省(区、市)共同富裕发展水平等级进行预测。研究结果表明:(1)2016—2022年,我国共同富裕发展水平从东至西呈现“高—中—低”阶梯式分布,但贫富差距显著缩小;(2)结合梯度提升算法研究我国共同富裕发展水平效果较好,其中XGBoost分类算法的预测结果最优。基于研究结论,本文分别从经济质量、协调发展、公共服务等方面提出政策建议,以期对推动共同富裕发展及后续相关研究提供参考。 展开更多
关键词 共同富裕 CRITIC权重法 K-Means++聚类算法 梯度提升算法
在线阅读 下载PDF
通过Grad-CAM++提升GIS局部放电深度诊断模型的泛化性评估精度
16
作者 闫泽玉 刘云鹏 +2 位作者 范晓舟 范澜珊 肖海 《中国电机工程学报》 北大核心 2025年第21期8622-8633,I0029,共13页
为提升现有局部放电深度诊断模型的泛化性评估精度和诊断结果的可解释性,该文提出样本特征提取效果和模型泛化能力的量化指标--对焦系数及其复合指标。通过改进的梯度权重类激活映射图(improved gradient-weighted class activation map... 为提升现有局部放电深度诊断模型的泛化性评估精度和诊断结果的可解释性,该文提出样本特征提取效果和模型泛化能力的量化指标--对焦系数及其复合指标。通过改进的梯度权重类激活映射图(improved gradient-weighted class activation mapping,Grad-CAM++)计算训练样本的类激活映射图(class activation mapping,CAM),并将其与局部放电相位分布(phase resolved partial discharge,PRPD)图卷积后除以调整因子,获得对焦系数。调整因子是将CAM按照局放次数进行最优分配后的卷积结果,是完美深度诊断模型的估计值。因此,对焦系数可以看作现有模型特征提取能力与理想模型的比值,其值大小在一定程度上能够表征模型的泛化能力。在此基础上,构建准确率与对焦系数的复合指标,进行最终模型的综合评估。通过110kV真型气体绝缘组合电器(gas insulated switchgear,GIS)平台收集7种多源缺陷局放数据,构建12个深度诊断模型。通过试验数据和现场数据构建6组测试数据集,验证该评估方法的有效性。结果表明,对焦系数能够有效量化CAM的可视化分析结果,提高诊断结果的置信度。综合对焦系数构建的复合指标(γ=10%)泛化相关性为81.01%,相较准确率指标提升8.74%。通过复合指标优选的诊断模型在现场陌生数据集下准确率为97%,相对传统优选方法准确率提升9.1%。 展开更多
关键词 局部放电 梯度加权类激活映射图 深度神经网络 可解释性 泛化性 气体绝缘组合电器
原文传递
基于夜间灯光数据的黄河流域城市建设用地碳排放影响因素分析及达峰预测
17
作者 薛建春 曹凌辰 +1 位作者 张翔 丁志斌 《沈阳农业大学学报》 北大核心 2025年第3期148-162,共15页
[目的]利用夜间灯光数据与省域CEADs碳排放构建拟合模型,计算黄河流域2000–2022年各地级市的城市建设用地碳排放。[方法]使用标准差椭圆以及重心模型分析探索黄河流域城市建设用地碳排放分布重心和重心移动趋势,通过全局莫兰指数和LIS... [目的]利用夜间灯光数据与省域CEADs碳排放构建拟合模型,计算黄河流域2000–2022年各地级市的城市建设用地碳排放。[方法]使用标准差椭圆以及重心模型分析探索黄河流域城市建设用地碳排放分布重心和重心移动趋势,通过全局莫兰指数和LISA集聚分析其空间相关性与高值聚集情况,最后利用GTWR模型从经济、人口、技术的角度分析各城市建设用地碳排放影响因素的回归系数水平,采用情景设定法探索不同情景下黄河流域建设用地“碳达峰”情况。[结果]黄河流域城市建设用地碳排放主要集中在下游、中游城市,但碳排放重心有向中游移动的趋势,坐标由111°82′81″E、37°01′675″N转移至111°56′46″E、36°53′425″N。同时碳排放在黄河流域内存在空间自相关性,下游地区部分城市存在高高集聚的现象。研究时期内,经济发展水平的回归系数逐渐减小,但在中游地区逐渐增加;人口规模的回归系数逐渐增大,说明由人口规模引发的碳排放量逐渐增大。[结论]经济高速发展情景下,建设用地“碳达峰”时间为2030年,基准与低碳发展情景下建设用地“碳达峰”时间均为2024年,峰值由大到小为:高速发展>基准>低碳情景。 展开更多
关键词 黄河流域 城市建设用地碳排放 影响因素 空间自相关模型 GTWR XGBoost
在线阅读 下载PDF
基于可解释深度学习及表面增强拉曼光谱的微塑料高效识别方法
18
作者 张艺严 马静 +1 位作者 孙振丽 杜晶晶 《分析测试学报》 北大核心 2025年第8期1557-1567,共11页
微塑料(MPs)污染已成为全球环境的重大挑战。传统检测方法在MPs检测中存在诸多局限,迫切需要开发无需复杂前处理的高灵敏检测技术。为解决MPs检测难题,该研究构建了一种“表面增强拉曼散射基底捕获-深度学习识别-梯度加权类激活映射(Gra... 微塑料(MPs)污染已成为全球环境的重大挑战。传统检测方法在MPs检测中存在诸多局限,迫切需要开发无需复杂前处理的高灵敏检测技术。为解决MPs检测难题,该研究构建了一种“表面增强拉曼散射基底捕获-深度学习识别-梯度加权类激活映射(Grad-CAM)解释”的MPs新型检测方法。研究结果表明,金纳米海绵基底可有效捕获MPs,数据增强与预处理技术可有效提高模型的预测精度。基于一维卷积神经网络(1D-CNN)的多分支二分类深度学习网络对MPs的分类准确率可达85%,显著高于机器学习模型与常规1DCNN模型。Grad-CAM分析可清晰展示模型决策依据及误判原因。该方法在实际环境水样及混合样本中验证有效,具有较强抗干扰性能与实用性;所用基底材料来源广泛、制备工艺简便,具有成本优势与良好的应用潜力。 展开更多
关键词 表面增强拉曼光谱 微塑料 卷积神经网络 深度学习 梯度加权类激活映射
在线阅读 下载PDF
基于改进SDP和FasterNet-GCAM的滚动轴承故障诊断
19
作者 陈家芳 唐湛恒 周健 《现代制造工程》 北大核心 2025年第7期129-138,41,共11页
对滚动轴承进行故障诊断关乎设备运行安全及稳定可靠性。使用传统卷积神经网络进行故障诊断,模型运算量过大,且易出现过拟合现象从而导致诊断精度不高,端到端模型存在可信度不高等问题。鉴于此,提出一种基于改进对称极坐标(Symmetrized ... 对滚动轴承进行故障诊断关乎设备运行安全及稳定可靠性。使用传统卷积神经网络进行故障诊断,模型运算量过大,且易出现过拟合现象从而导致诊断精度不高,端到端模型存在可信度不高等问题。鉴于此,提出一种基于改进对称极坐标(Symmetrized Dot Pattern,SDP)法和FasterNet-GCAM网络的滚动轴承故障诊断方法。首先,将一维振动信号经过小波阈值降噪处理,再输入经皮尔逊图像相关系数法优化的SDP法生成SDP图像,并通过在FasterNet网络中加入部分卷积(partial convolution)思想,构建成改进的SDP-FasterNet模型进行进一步的特征提取,并完成滚动轴承不同故障的分类诊断。为了验证模型在图像识别过程中决策的可信度,将梯度加权类激活映射(Gradient-weighted Class Activation Mapping,Grad-CAM)与FasterNet网络相结合,突出SDP图像与决策相关的重要部分。试验结果表明,所提方法相比于其他方法具有更快的收敛速度和更强的鲁棒性,且诊断识别精度达到了99.20%,并提高了诊断过程中的可解释性及可信度,为故障诊断领域提供了具备良好可行性和鲁棒性的轻量化诊断模型。 展开更多
关键词 滚动轴承 故障诊断 FasterNet网络 部分卷积 梯度加权类激活映射 对称极坐标法
在线阅读 下载PDF
基于梯度权值追踪的域自适应分类研究
20
作者 崔绍君 季繁繁 +1 位作者 王婷 袁晓彤 《南京信息工程大学学报》 北大核心 2025年第2期203-214,共12页
本文提出一种基于梯度权值追踪的剪枝与优化算法(GWP),旨在解决无监督领域中存在的过拟合问题,即在下游任务上的精度远低于在训练集上的精度.针对无监督领域自适应中基于差异与基于对抗的方法,将稠密-稀疏-稠密策略应用于解决过拟合问题... 本文提出一种基于梯度权值追踪的剪枝与优化算法(GWP),旨在解决无监督领域中存在的过拟合问题,即在下游任务上的精度远低于在训练集上的精度.针对无监督领域自适应中基于差异与基于对抗的方法,将稠密-稀疏-稠密策略应用于解决过拟合问题.先对网络进行密集预训练,并学出哪些连接是重要的;在剪枝阶段,与原有的稠密-稀疏-稠密策略中的剪枝过程不同,本文的优化算法同时将权值和梯度联合考虑,既考虑到了权值信息(即零阶信息),也考虑到了梯度信息(即一阶信息)对网络剪枝过程的影响;在重密集阶段,恢复被修剪的连接,并以较小的学习率重新训练密集网络.最终,得到的网络在下游任务上取得了理想的效果.实验结果表明,与原有的基于差异和基于对抗的领域自适应方法相比,本文提出的GWP可以有效提升下游任务精度,且具有即插即用的效果. 展开更多
关键词 梯度权值追踪 无监督领域自适应 稠密-稀疏-稠密 过拟合 零阶信息 一阶信息
在线阅读 下载PDF
上一页 1 2 23 下一页 到第
使用帮助 返回顶部