Active features extracted by deep belief network for process monitoring

Recently, based on the powerful capability of feature extraction, deep learning technique has been applied to the field of process monitoring, and usually, the researches utilize all the abstract features to establish the detection model and detect or classify the fault. However, whether all the extracted features are valid and beneficial for process monitoring have never been researched and discussed. If there are some features that are adverse for process monitoring, the detection performance of the model would be reduced once they are considered in the model, and utilized the features that are advantageous for process monitoring could ameliorate the performance of detection model. Motivated by this, a feasibility analysis on each feature captured by deep belief network for process monitoring is executed and the conception of active features (AFs) which have active expression for the occurrence of the fault is proposed. Based on AFs, utilized Euclidean metric to calculate the dissimilarity between the test sample and the training sample, and moving average technique is employed to reduce the effect of the burst noise in measurement variables on the result. Finally, the comparison of fault detection rate with other advanced methods on a numerical process and TE process demonstrate the feasibility and superiority of the proposed method, AF-DBN in this study.     

Validating the moderating role of age in multi-perspective acceptance model of wearable healthcare technology

The recent trend of aging population, not to mention the unprecedented pandemic, draws great attention from the general public about health concerns. Since healthcare information technology is different from non-healthcare information technology, additional contexts should be properly incorporated into technology acceptance research to accurately identify influential factors affecting the acceptance of wearable healthcare technology. Thus, we selectively reconfigured factors from health, privacy, and socio-demographic contexts to formulate a health-aware acceptance model. Then, it was empirically analyzed using structural equation modeling. Based on the results, whereas privacy concerns were directly associated with intention to use wearable healthcare technology, health concerns were not. Moreover, age had a moderating effect on social influence and facilitating conditions. These findings suggest valuable insights that the adoption rate of healthcare technology is increased by 1) keeping personal information securely, 2) facilitating social interaction among users, and 3) offering intuitive user experience for elderly people.     

基于深度信念网络的列车故障音频诊断方法

铁路在交通运输行业有着举足轻重的地位，一旦列车发生故障将会导致严重的生命财产损失。由于列车发生故障的概率相对较低，因此难以捕获列车的故障样本。针对上述问题，提出了一种无监督学习的列车故障识别方法，通过检测列车音频信号来识别列车故障。该方法基于深度信念网络(DBN)，利用小波包分解提取检测信号的特征向量并将其作为DBN的输入，待网络充分训练后，由训练好的DBN识别当前列车的运行状况。现场监测实验结果表明，该方法能够在无监督的条件下有效识别列车故障，保障了列车的运行安全。     

基于Adam-DBN网络的行星齿轮箱故障诊断方法研究

随着现代机械装备的复杂化,传统的故障诊断方法难以满足表征设备间的复杂映射关系,且在如今大数据背景下面临着维数灾难的问题。文章结合深度置信网络以及梯度优化算法提出了一种基于梯度优化深度置信网络(Adam-DBN),通过数据验证选取最优梯度优化算法对深度置信网络的的梯度算法进行调优。搭建模拟实际工况的行星齿轮箱实验台,通过实验台采集所得数据构成数据集对方法进行验证。实验表明文中提出的方法能够有效提高DBN网络的收敛速度与训练精度,同时具有较高的故障识别准确率。     

基于改进深度信念网络模型的中长期径流预测

为提高流域中长期径流预测效果,提出径流综合指数构建、因子筛选和改进深度信念网络模型相结合的预测方法。首先研究不同水文站点（细粒度）月平均径流的一致性,构造流域径流综合指数（粗粒度）,在较宏观层面研究流域水情丰枯变化;接着采用基于信息熵的因子筛选方法,获得影响流域水情丰枯变化的关键因子集,形成深度学习的输入;然后利用改进的深度信念网络（IDBN）模型进行预测。以雅砻江流域为例,将所建模型与多元线性回归、自回归移动平均、反向传播（BP）神经网络、支持向量机和传统深度信念网络等预测模型进行对比分析。结果表明：所提方法具有较好的实用性,且IDBN模型具有更好的预测速度和精度。研究结果可为流域中长期径流变化趋势预测提供参考。     

基于深度信念网络的低风速风电参与微电网频率优化控制

低风速分散式风电接入微电网使得微电网能够利用低风速风资源为负荷供电,但其低惯量的特点也对微电网的频率稳定性提出了挑战。为了探究微电网中低风速风电机组(LWTG)如何有效参与抑制微电网频率波动,在LWTG中引入虚拟惯量控制、超速控制和下垂控制。针对风电机组最小转子转速限制,通过理论分析确定了合适的LWTG的参数;针对超速控制存在的盲区问题,利用深度信念网络来优化不同风速下的减载率以及虚拟惯量控制、下垂控制的控制参数;在低风速风况下验证了优化后的参数能够有效减少负荷波动引起的微电网动态频率跌落幅度,并获得较好的调频效果。     

基于深度信念网络的计量装置故障溯源研究

针对电网海量大数据中存在异常的电表数据,提出了一种基于深度信念网络的计量装置故障溯源模型。首先,分析了深度信念网络（DBN）模型基本原理,提出了一种智能电表故障分类DBN结构模型,并给出了计量装置故障溯源建模流程;然后,通过建立离线台账样本库、实时用电曲线故障特征样本库,实现了计量装置故障样本库构建;最后,基于实际电表测试和数据异常识别,完成了计量装置的台账样本库溯源测试,并与已有的贝叶斯、决策树、随机森林、决策树提升等分类算法进行比较,测试结果验证了所提模型和方法的正确性和有效性。     

A fuzzy Bayesian belief network for safety assessment of oil and gas pipelines

Safety assessment of oil and gas (O&G) pipelines is necessary to prevent unwanted events that may cause catastrophic accidents and heavy financial losses. This study develops a safety assessment model for O&G pipeline failure by incorporating fuzzy logic into Bayesian belief network. Proposed fuzzy Bayesian belief network (FBBN) model explicitly represents dependencies of events, updating probabilities and representation of uncertain knowledge (such as randomness, vagueness and ignorance). The study highlights the utility of FBBN in safety analysis of O&G pipeline because of its flexible structure, allowing it to fit a wide variety of accident scenarios. The sensitivity analysis of the proposed model indicates that construction defect, overload, mechanical damage, bad installation and quality of worker are the most significant causes for the O&G pipeline failures. The research results can help owners of transmission and distribution pipeline companies and professionals to prepare preventive safety measures and allocate proper resources.     

网上试卷分析技术的研究与实现

所介绍的“网上试卷分析技术的研究与实现”是考试系统的一个子功能 .试卷分析是考试系统效果的一个重要保证 ,目前在网络教育方面有关试卷分析的研究极少 .文章中主要论述了在系统实现时 ,如何应用统计学中的信度、效度估计方法 ,对试卷的信度、效度、区分度进行估计 .并介绍了系统具体实现时采用的 MVC模式设计和开发过程中的应用技术 .