基于FUZZY-BN-FTA的厂区架空燃气管道泄漏可能性研究

为研究厂区架空燃气管道泄漏的故的可能性,提出模糊数学方法、贝叶斯网络模型以及故障树模型相结合的集成模型。通过GeNIe软件建立架空燃气管道泄漏事故的模糊贝叶斯网络模型;引入Leaky Noisy-or Gate扩展模型对模糊贝叶斯网络中节点的条件概率表进行修正,结合基本事件先验概率预测架空燃气管道泄漏的概率值,并引入3个基本事件重要度确定基本事件的重要性排序,同时进行敏感性分析确定主要影响因素,提出控制措施。研究结果表明:厂区车辆撞击、管道疲劳损耗、管材质量不佳、安装设计不合理、焊接缺陷为导致架空燃气管道泄漏的主要因素。研究结果可为工业企业安全管理人员有针对性地制定控制措施,减少事故发生率提供参考。

2-O-β-D-Glucopyranosyl-L-ascorbic acid,an ascorbic acid derivative isolated from the fruits of Lycium barbarum L.,ameliorates high fructose-induced neuroinflammation in mice:involvement of gut microbiota and leaky gut

Western diet(rich in highly refined sugar and fat)can induce a range of metabolic dysfunctions in animals and humans,including neuroinflammation and cognitive function decline.Neuroinflammation and cognitive impairment,two critical pathological characteristics of Alzheimer’s disease,have been closely associated with microbial alteration via the gut-brain axis.Thus,the present study aimed to investigate the influence of 2-O-β-D-glucopyranosyl-L-ascorbic acid(AA-2βG)isolated from the fruits of Lycium barbarum on preventing the high-fructose diet(HFrD)induced neuroinflammation in mice.It was found that AA-2βG prevented HFr D-induced cognitive deficits.AA-2βG also predominantly enhanced the gut barrier integrity,decreased lipopolysaccharide entry into the circulation,which subsequently countered the activation of glial cells and neuroinflammatory response.These beneficial effects were transmissible by horizontal fecal microbiome transplantation,transferring from AA-2βG fed mice to HFr D fed mice.Additionally,AA-2βG exerted neuroprotective effects involving the enrichment of Lactobacillus and Akkermansia,potentially beneficial intestinal bacteria.The present study provided the evidence that AA-2βG could improve indices of cognition and neuroinflammmation via modulating gut dybiosis and preventing leaky gut.As a potential functional food ingredient,AA-2βG may be applied to attenuate neuroinflammation associated with Western-style diets.

Presepsin as a biomarker of bacterial translocation and an indicator for the prescription of probiotics in cirrhosis

BACKGROUND The gut–liver axis and bacterial translocation are important in cirrhosis,but there is no available universal biomarker of cellular bacterial translocation,for which presepsin may be a candidate.AIM To evaluate the relationship of the blood presepsin levels with the state of the gut microbiota in cirrhosis in the absence of obvious infection.METHODS This study included 48 patients with Child–Pugh cirrhosis classes B and C and 15 healthy controls.The fecal microbiome was assessed using 16S rRNA gene sequencing.Plasma levels of presepsin were measured.A total of 22 patients received a probiotic(Saccharomyces boulardii)for 3 months.RESULTS Presepsin levels were higher in patients with cirrhosis than in healthy individuals[342(91-2875)vs 120(102-141)pg/mL;P=0.048].Patients with elevated presepsin levels accounted for 56.3%of all included patients.They had lower levels of serum albumin and higher levels of serum total bilirubin and overall severity of cirrhosis as assessed using the Child–Pugh scale.Patients with elevated presepsin levels had an increased abundance of the main taxa responsible for bacterial translocation,namely Bacilli and Proteobacteria(including the main class Gammaproteobacteria and the minor taxa Xanthobacteraceae and Stenotrophomonas),and a low abundance of bacteria from the family Lachnospiraceae(including the minor genus Fusicatenibacter),which produce short-chain fatty acids that have a positive effect on intestinal barrier function.The presepsin level directly correlated with the relative abundance of Bacilli,Proteobacteria,and inversely correlated with the abundance of Lachnospiraceae and Propionibacteriaceae.After 3 months of taking the probiotic,the severity of cirrhosis on the Child–Pugh scale decreased significantly only in the group with elevated presepsin levels[from 9(8-11)to 7(6-9);P=0.004],while there were no significant changes in the group with normal presepsin levels[from 8(7-8)to 7(6-8);P=0.123].A high level of presepsin before the prescription of the probiotic was an independent predictor of a greater decrease in Child–Pugh scores(P=0.046),as well as a higher level of the Child–Pugh scale(P=0.042),but not the C-reactive protein level(P=0.679)according to multivariate linear regression analysis.CONCLUSION The level of presepsin directly correlates with the abundance in the gut microbiota of the main taxa that are substrates of bacterial translocation in cirrhosis.This biomarker,in the absence of obvious infection,seems important for assessing the state of the gut–liver axis in cirrhosis and deciding on therapy targeted at the gut microbiota in this disease.

基于改进VGG16图像分类方法研究

针对神经网络模型在训练过程中遇到的收敛速度慢和测试样本不平衡导致的准确率降低问题,提出了一种基于改进VGG16图像分类模型的LBF-VGG16(Leaky-Bactch-Focal-VGG16)。该模型将原Relu激活函数替换为Leaky Relu,并在卷积层与激活函数之间引入BN层,以优化收敛效果。在训练过程中,采用SGD优化器,并融入Focal Loss损失函数。试验结果表明,LBF-VGG16模型在分类效果和收敛速度方面较改进前均有显著提升。

基于改进卷积神经网络的辣椒病虫害检测

针对使用卷积神经网络对辣椒病虫害进行检测有参数多、计算量大和推理时间过长等问题,提出一种基于MobileNet-V2改进的轻量化神经网络,将MobileNet-V2的BN层中的激活函数全部替换为Leaky ReLU,保留特征图中更多的有效正负信息,以提高性能和减少计算复杂度,增强模型的鲁棒性。在公开的辣椒病虫害数据集上使用VGG16、ResNet34和MobileNet-V2等模型对比后,改进的MobileNet-V2表现出更高的准确性和更少的参数量。相对于原来的MobileNet-V2准确率提升4%,相对VGG16、ResNet34两种模型参数分别下降97%和87%。能够移动端设备实现实时病虫害检测,提供高效便捷解决方案。

基于图像增强和注意力机制的作物杂草识别

为提高复杂环境下无人机获取的作物杂草图像识别的准确率,提出一种基于图像增强与注意力机制的作物杂草识别方法。在多尺度Retinex算法中加入颜色恢复函数调节3个通道颜色的占比以恢复其颜色特征,使图像更清晰;将残差网络模型中的激活函数换为Leaky ReLU,加入CBAM注意力机制模块,获取更多有用信息,抑制其它无用信息。实验结果表明,该方法可以提高复杂环境下无人机获取的作物杂草图像的识别准确率,其准确率达到95.3%,高于AlexNet、ResNet18、ResNet50及其它主流算法的识别结果。

基于改进CNN的苹果缺陷检测方法研究

目的:解决现有苹果缺陷检测方法存在的精度低、效率差等问题。方法:基于水果图像采集系统,提出一种改进的卷积神经网络用于苹果表面缺陷检测;引入深度可分离卷积代换原网络标准卷积,提高特征提取速度;引入Leaky ReLU激活函数代替ReLU激活函数,提高计算效率和精度;引入全局平均池化替换全连接层,降低网络模型的计算量;并在每层卷积后加入批量归一化层,通过试验与常规方法进行对比分析,验证其优越性。结果:与常规方法相比,所提方法在苹果缺陷检测中具有较高的检测准确率和速度,且模型参数量少,准确率达99.60%,检测速度(每秒帧数)达526,模型参数量为389 072。结论:该苹果缺陷检测方法能有效降低模型参数和检测时间,具有较高的准确率和速度。

基于Leaky Noisy-or Gate和贝叶斯网络的光伏发电项目施工风险评估方法

针对复杂多变的光伏发电项目施工风险评估问题,利用Leaky Noisy-or Gate模型整合历史经验和专家知识来确定贝叶斯网络参数并建立光伏施工风险的贝叶斯网络模型,同时采用Netica仿真分析,对施工风险进行有效预测并诊断识别出关键性风险和敏感性风险,以便管理者精准施策、管控风险。为验证该方法的有效性,以Z项目为例进行建模仿真。结果表明,其施工风险等级为中等,与实际相符。

基于有效通道注意力EfficientNet的肝硬化识别

针对传统卷积神经网络模型复杂度高、参数量大,网络分类的精度和效率不佳等问题,提出一种应用于医学超声图像中肝硬化识别的深度学习方法(E-EfficientNet)。在EfficientNet网络模型中,将模块MBConv中的注意力机制SENet模块替换为一种不降维的ECANet模块,避免降维操作导致的特征信息缺失,增强通道学习能力并降低模型复杂度;将可变形卷积融入EfficientNet网络,利用可变形卷积核能够依据目标形态自适应调整变化的特点,充分学习图像细节,提升算法的泛化能力和特征提取能力;对有限样本进行数据增强,避免训练过程中出现过拟合,并使用Leaky ReLU作为激活函数保留图像负值特征信息,提高对肝硬化的识别效果。实验结果表明,该模型复杂度低,在肝硬化识别中准确率为98.9%。

Leaky Cable Fixture Detection in Railway Tunnel Based on RW DCGAN and Compressed GS-YOLOv5

The communication system of high-speed trains in railway tunnels needs to be built with leaky cables fixed on the tunnel wall with special fixtures.To ensure safety,checking the regular leaky cable fixture is necessary to elimi-nate the potential danger.At present,the existing fixture detection algorithms are difficult to take into account detection accuracy and speed at the same time.The faulty fixture is also insufficient and difficult to obtain,seriously affecting the model detection effect.To solve these problems,an innovative detection method is proposed in this paper.Firstly,we presented the Res-Net and Wasserstein-Deep Convolution GAN(RW-DCGAN)to implement data augmentation,which can enable the faulty fixture to export more high-quality and irregular images.Secondly,we proposed the Ghost SENet-YOLOv5(GS-YOLOv5)to enhance the expression of fixture feature,and further improve the detection accuracy and speed.Finally,we adopted the model compression strategy to prune redundant channels,and visualized training details with Grad-CAM to verify the reliability of our model.Experimental results show that the algorithm model is 69.06%smaller than the original YOLOv5 model,with 70.07%fewer parameters,2.1%higher accuracy and 14.82 fps faster speed,meeting the needs of tunnel fixture detection.

A Survey on Channel Measurements and Models for Underground MIMO Communication Systems

The high reliability of the communication system is critical in metro and mining applications for personal safety,channel optimization,and improving operational performance.This paper surveys the progress of wireless communication systems in underground environments such as tunnels and mines from 1920 to 2022,including the evolution of primitive technology,advancements in channel modelling,and realization of various wireless propagation channels.In addition,the existing and advanced channel modeling strategies,which include the evolution of different technologies and their applications;mathematical,analytical,and experimental techniques for radio propagation;and significance of the radiation characteristics,antenna placement,and physical environment of multiple-input multiple-output(MIMO)communication systems,are analyzed.The given study introduces leaky coaxial cable(LCX)and distributed antenna system(DAS)designs for improving narrowband and wideband channel capacity.The paper concludes by figuring out open research areas for the future technologies.

浅谈基于贝叶斯网络的泵故障诊断方法

针对发电厂水泵系统结构复杂,诊断过程中不确定因素众多的特点,提出了基于贝叶斯网络的水泵故障诊断方法。依据发电厂水泵故障机理,建立水泵贝叶斯网络结构;在定量参数方面,引入leaky Noisy-or模型,简化参数设置,从而克服水泵故障样本数量少导致贝叶斯网络参数设置困难的问题。最后,通过实例验证了基于贝叶斯网络的发电厂水泵故障诊断方法的有效性。

用于态势估计的一种构造贝叶斯网络参数的方法

根据态势估计的特点和要求,采用Leaky noisy-or gate模型的方法构造,并将其应用于整个贝叶斯网络中。该方法以离散变量集为研究对象,由于满足构成Leaky noisy-or gate模型的条件,能够利用部分统计信息构造节点的网络参数。文中利用相关的贝叶斯网络推理软件进行了实验。实验结果表明,使用Leaky noisy-or gate模型的方法,对网络参数进行构造是可行的。

外电场作用下改进的Leaky Integrate-And-Fire模型的峰放电频率适应性研究

峰放电频率适应性是神经元在信息处理过程中重要的动力学特性之一.当神经系统受到外电场作用时,会对其动力学行为以及神经电信息的产生、传导产生影响.我们基于Leaky integrate-and-fire(LIF)神经元模型,建立了外电场作用下改进的LIF神经元模型.采用随时间演化的膜电位曲线和峰放电频率曲线,以及随外电场变化的起始峰放电频率曲线和稳态峰放电频率曲线,研究不同强度、频率外电场作用下改进的LIF模型的适应性变化.此外,还利用相邻峰峰间期(ISI)之间的相关性进一步阐明外电场对神经元适应性的影响.

信息不完备情况下多因素工序质量诊断方法

为解决信息不完备情况下的多因素工序质量诊断问题,在工艺机理分析的基础上,提出基于贝叶斯网络模型构建与推理的问题溯源方法.在贝叶斯网络结构学习过程中,利用基于评分/搜索的思想对基于工艺的预先假设结构,通过互信息参量排序降低学习复杂度.针对生产过程中随机因素对诊断准确性的影响问题,结合Leaky Noisy-OR模型引入随机参量节点,对数据需求和推理进行降解优化.以沟道磨削表面形貌质量问题的诊断为例,给出模型构建与推理程序,并验证了所构建模型及优化方法的可行性和有效性.

基于贝叶斯网络的空调冷水阀故障检测与诊断

针对空调系统运行中所产生的很多不确定性故障问题,建立了贝叶斯网络模型,开发了多故障状态的贝叶斯概率计算模型,利用Leaky Noisy-Max模型进行贝叶斯网络参数学习,提出了基于故障特征准则和阈值判别的故障检测与诊断技术。采用现场实测数据,以冷水阀故障为例,验证了该技术的可行性。结果表明,该技术能够诊断出冷水阀故障,诊断结果能够快速定位故障产生源,从而提高空调运行系统的稳定性。

基于改进深度残差网络的低功耗表情识别

为了提高表情识别率并降低表情识别的功耗,提出一种基于改进深度残差网络的表情识别方法。残差学习在解决深度卷积神经网络退化问题、使网络层次大幅加深的同时,进一步增加了网络的功耗。为此,引入具有生物真实性的激活函数来代替已有的整流线性单元(Rectified Linear Units,ReLU)函数,并将其作为卷积层激活函数对深度残差网络进行改进。该方法不仅提高了残差网络的精度,而且训练出的网络权重可直接作为与该深度残差网络具有相同结构的深度脉冲神经网络的权重。将该深度脉冲神经网络部署在类脑硬件上时,其能够以较高的识别率和较低的能耗进行表情识别。

“一带一路”基础设施PPP项目投资风险研究——以东南亚三国为例

基于国内投资方视角,针对“一带一路”基础设施PPP项目投资风险问题,以投资高风险的东南亚三国(柬埔寨、老挝、缅甸)为例,通过调查问卷获得投资风险相关数据,确定高等级风险:政府腐败/不作为、政府信用不足。其次利用贝叶斯网络建立“一带一路”基础设施PPP项目投资风险网络模型,使用改进的Leaky noisy-or gate模型完善网络参数。采用团树传播算法获得东南亚三国投资关键性风险:市场收益不足、外汇风险、政府腐败/不作为,运用GeNIe 2.0软件获得敏感性风险:违背PPP国际规则。并给出“一带一路”基础设施PPP项目投资风险应对措施和投资建议。

基于多特征融合CNN的人脸识别算法研究

针对传统卷积网络结构或使用单一特征融合方法进行人脸识别存在特征提取不全、训练准确率低的问题,提出一种基于多特征融合卷积神经网络(CNN)的人脸识别算法。算法在LeNet5结构基础上,通过不同卷积层映射图融合、多个特征核映射图融合和浅层纹理的轮廓特征和深层高级特征相融合,增加输出通道数,进而提高图像整体的语义信息。针对relu激活函数在神经网络训练过程中丢失负轴信息的缺陷,算法引入leaky relu激活函数。通过在ORL、AR、FERET三个人脸库上分别对LeNet5卷积神经网络结构、单一特征融合与算法进行对比实验。实验结果表明:算法通过多特征融合提取的特征信息更广,识别准确率高于单一特征融合,引入leaky relu激活函数后,网络收敛效果更好,同时,对遮挡、光照等干扰具有鲁棒性。

基于卷积神经网络的磁瓦表面缺陷识别

针对现有的磁瓦表面缺陷识别算法准确率低且泛化能力不足的问题,提出一种基于卷积神经网络的磁瓦表面缺陷识别算法。该算法在轻量级NASNet卷积神经网络的基础上,引入双线性模型增强网络的特征表达力,提高算法识别精度。其次,为了减少模型参数量与计算量,改进NASNet模型结构进行特征降维。同时采用Adam算法作为网络的优化算法,提高训练过程的平稳性和收敛速度;选择Leaky ReLU函数作为网络的激活函数,提高模型的鲁棒性。考虑小样本磁瓦数据集,引入Dropout、样本增强、迁移学习技术防止网络过拟合,提高网络模型的泛化能力。实验结果表明,该算法实现了小样本多类别磁瓦表面缺陷的高精度识别,平均识别精度达到98.8%,与轻量级NASNet相比平均识别率提升约2.49%。