利用小波变换检测心电图ST段

针对心电图ST段检测精确度不高的问题,依据小波变换理论,提出了一种心电图ST段检测方法,检测心电图的特征值来准确确定J点及T波起点,并结合传统J＋X法及T波起点检测方法达到准确检测ST段电位的目的。该方法通过MIT-BIH的STDB数据库和European ST-T数据库进行检验,ST段检测误差在4ms内的占94．5％,在8ms的占98．9％,提高了检测精确度。     

考虑水平和竖直方向振动作用的导线疲劳损伤检测方法研究

针对架空导线振动损伤检测方法计算精度不高的问题，提出一种更为精确的导线损伤检测方法。该方法以不同风速和张力下具有水平和竖直动力的不对称双跨导线结构响应特征为基础，研究水平和竖直方向振动作用的导线疲劳损伤程度。通过薛氏模型识别出水平和竖直加速度信号，利用深度孪生神经网络(deep Siamese convolutional network, DSCN)架构将水平和竖直方向加速度信号输入标准化时间窗，然后经过希尔伯特黄变换-支持向量机(Hilbert Huang transform-support vector machine, HHT-SVM)算法求解导线疲劳损伤值，并与主流算法对比验证。不同风速条件下的数据结果显示，HHT-SVM方法较遗传算法、卷积神经网络(convolutional neural network, CNN)算法两种方法更为优异，计算的疲劳损伤值总体精度分别提高了22.77%、8.06%,模态阶数精度提高了23%、4%,迭代次数减少了45%、17%。最后，通过HHT-SVM方法，结合三维颤振稳定性检测图，得出导线振动与疲劳损伤值的关系与推导出的比例关系一致，从而再次验...     

基于特征量优选与ICA-SVM的变压器故障诊断模型

为了弥补现有变压器故障诊断方法在油中气体分析(DGA)特征量选取和诊断模型方面的不足,采用IEC三比值法中的3种气体比值作为变压器故障诊断的特征量。同时从含有8种油中溶解气体中任意3种及以上的共254种气体组合中筛选出准确率最高的3组最优DGA特征气体组合,将其作为对照组特征量。然后采用帝国竞争算法(ICA)优化支持向量机的变压器故障诊断模型(ICA-SVM),与标准支持向量机(SVM)法、粒子群优化向量机(PSO-SVM)以及IEC三比值法进行对比。实例结果表明:三气体比值特征量相比3组最优DGA气体组合,故障识别准确率提高了10%左右;ICA-SVM故障诊断模型相比标准SVM法、PSO-SVM和IEC三比值法故障识别准确率提高了7%～35%;综合三比值特征量与ICA-SVM故障诊断模型的准确率为89.3%,相较其他几种方法准确率提升了7%～35%。结果验证了该方法的有效性和准确性。     

心电信号的卷积神经网络二分类方法

心电信号直观地反映心脏生理电活动,在诊断心血管疾病方面有重要的参考价值。提出了一种卷积神经网络的心电信号二分类方法,网络卷积层使用不同卷积核,最大限度的利用局部特征进行分类,对异常心拍进行检测。使用麻省理工学院提出的MIT-BIH心律失常数据库对该方法进行验证。通过混淆矩阵计算性能指标,运用交叉验证与3种传统机器学习方法对比。实验表明,相较于准确性能最高的支持向量机二分类方法,模型准确率可达96.86%,提升了3.39%。该方法简化了特征提取过程,充分提高了异常心拍检测的准确性。     

基于ST7MC的无位置传感器BLDC反电势过零检测

本文通过对梯形波反向电动势和绕组电路模型的分析,具体阐明了反电势过零检测方法的原理,提出一种基于ST7MC的反电势过零检测设计方案,该方案采用上臂PWM,下臂恒通的调制方式,在PWM的OFF阶段进行反电势过零检测。与传统方案比较,该方案具有硬件成本低、抗噪性能好、能对反向电动势进行无衰减的检测,以及扩大了电机的速度范围等优点。     

基于梯度提升树的电压暂降源概率辨识

提出一种基于梯度提升树的电压暂降源概率辨识方法,通过多个决策树弱学习器的依次学习不断拟合残差,依据算法输出的概率结果对暂降源进行辨识。依据仿真获得的各类暂降源波形对该方法的有效性和准确性进行验证,并与传统的支持向量机算法进行了对比。在相同样本数量下该方法相较支持向量机具有更高辨识准确性,所给出的各类暂降源辨识概率信息更能反映模型辨识的可信度,利于辅助决策人员进行决策。     

基于ARM的开关柜环境湿度检测及除湿系统

针对传统开关柜、接线端子以及电子元器件容易受环境影响,潮湿的空气进入柜体接触到电子设备形成凝露,降低了绝缘性能,造成事故,同时不能实时智能控制处理,造成资源浪费。设计了基于ARM的新型开关柜环境湿度检测及除湿系统。该系统以STM32为核心,利用湿度传感器网络检测湿度参数,通过RS485传递参数数据,上位机实时显示数据,设置阈值参数,控制中心作出除湿处理。该系统通过4组数据与高精度湿度检测仪器对比,最小相对误差为0.06%,最大相对误差0.1%,准确度为0.1。能够真实呈现开关柜内的湿度状况。在48 min内,根据无有本除湿系统对比,湿度分别下降到67.2%和50%。从实验结果来看,系统准确度高,实时准确显示柜内湿度,智能除湿效果明显,能很好地检测处理开关柜环境湿度。     

基于灰色支持向量机的输电线路覆冰厚度预测模型

为了降低输电线路覆冰事故对电网安全造成的严重影响,对输电线路覆冰厚度进行预测将能够有效地指导电网抗冰工作。提出了基于灰色支持向量机的输电线路覆冰厚度短期预测模型,分析了样本中脏数据的剔除及数据预处理方法,通过模型预测值与实测数据的对比验证了该模型的准确性和适用性,根据模型预测的线路最大覆冰厚度值对现场观冰、冰情预警以及开展交直流融冰提供策略指导。将该模型与传统的支持向量机和广义回归神经网络覆冰预测模型进行了对比,结果表明,该模型平均误差为0.325 mm,平均绝对百分误差仅为2.61%,适用于输电线路覆冰厚度短期预测。在易覆冰地区,应用该预测模型能够更好地指导输电线路抗冰工作。     

基于支持向量机的铝板表面缺陷检测分类

论述了基于机器视觉的铝板表面缺陷检测方法,提取7种铝板表面缺陷的多种特征值作为训练数据进行学习。介绍了支持向量机的原理和核函数的选择,在VC环境中构建支持向量机分类模型,用该模型对铝板表面缺陷进行分类标识,最后对支持向量机分类器的分类准确性、稳健性作出评价。     

结合区域生长与水平集算法的宫颈癌图像分割

针对宫颈图像病灶分割时的初始轮廓敏感问题和图像灰度不明显问题,提出一种新的改进的水平集算法。 首先利用各 向异性滤波算法等进行图像的去噪;然后在二值图像上使用区域生长算法,提取出粗糙的宫颈病灶区域;最后建立一种基于新 的符号压力函数的水平集模型,对初始分割进行细化。 该算法可以将局部信息与全局信息结合起来并自动分配局部信息与全 局信息的比例。 以 3 种统计指标为标准对该方法进行了评估,该方法在准确性、敏感性和特异性上可分别达到 81. 11%、 63. 97%和 78. 64%,分别比传统水平集算法高 30. 69%、15. 15%和 4. 37%。 因而,这种改进的水平集算法在实际应用中有一定 的价值和意义。     

Heart Rate Variability in Myocardial Ischemic Periods

Accurate assessment of myocardial ischemia plays an important role in the diagnosis and the treatment of ischemic heart diseases. ST-segment deviation in the electrocardiogram (ECG), resulting from the acute changes in the repolarization of the heart electric signal due to ischemia, is a routine method for the diagnosis of myocardial ischemia.However, this kind ofmorphological detection presents some problems that reduce the sensitivity and the specificity of the detection. Such phenomena, as a change in heart rate or even a simple posture change, can cause similar manifestations in the ST segment [1], [2]. In fact, besides the electrical alternation, myocardial ischemia also induces the generation and inhibition of cardiac reflex mediated by the autonomic nerve system (ANS) [3]. Investigating the dynamic ANS activity variations induced by ischemia may be helpful in the detection of ischemic episode and in the understanding of ANS activity during ischemia.     

Detecting myocardial ischemia with 2-D spectrum analysis of VCGsignals

Based on the AFFT algorithm, the 2-D spectrum of the VCG signal method was applied in analyzing ECG signals for different extents of myocardial ischemia. Figures and parameters of the 2-D spectrum clearly reflect the vector-changing characteristics of the VCG signals in the 2-D plane and also in the frequency domain. This method remedies defects of the DFT spectrum, which cannot describe vector characteristics of the VCG signal on the 2-D plane. Through analysis of animal experiments, the authors find that the AFFT method is sensitive to change in myocardial blood flow. This method shows promise in the application of detecting early myocardial ischemia and measuring its extent. As a result, it is hoped that this method will significantly reduce patient mortality and morbidity     

Time-frequency analysis of heart-rate variability

We present the results of a study that shows the viability of a new technique for the diagnosis and monitoring of myocardial ischemia that is based on the utilization of heart-rate variability (HRV) information. Ischemia is understood as being the lack of oxygen supply to the heart, a situation that in an extreme and irreversible case results in acute myocardial infarction (AMI), a reason for which early detection and treatment is of great interest. The treatment of ischemia can be approached via the evolution of the ECG, and especially from one of the parameters extracted from it-the ST segment (ECG signal between S and T waves) deviation. The utility of this measure is found in its capacity for detecting abnormalities in the conduction of the cardiac impulse that are associated with the presence of ischemia     

A Bayesian approach for dealing with uncertainties in detection ofcoronary artery stenosis using a knowledge-based system

A knowledge-based system that combines subjective Bayesian methods with rules specified by cardiologists to diagnose coronary artery stenosis from postexercise myocardial perfusion scintigrams is discussed. This expert system was used to determine which of the three main coronary arteries had the dominant stenosis. The system also indicated when a patient had a normal myocardial perfusion pattern (no stenosis). The system was run on a set of scans from 91 patients, and the results were compared with an existing expert system that uses the Dempster-Shafer theory of evidence for dealing with uncertainties. The system was able to determine the coronary artery with the dominant stenosis over 90% of the time when supplied with prior knowledge that all the patients have single-vessel stenosis. The system was also able to determine with good accuracy whether a patient had a stenosed coronary artery or normal myocardial perfusion when no prior information was available. The program can be used initially to screen out patients with normal scintigrams. Once the patients with normal scintigrams have been removed, the expert system can then be run on the remaining patients and utilize prior knowledge that they have stenosed coronary arteries. This improves the reliability of the diagnosis.     

Application of GPT to tanδ measuring apparatus fordistribution cable in hot line

A hot line tanδ measuring apparatus which uses the secondary voltage of GPTs (grounding potential transformers) existing at substations and other facilities has been developed. An error vector circle diagram method which allows one to determine the value of tanδ to an accuracy of 0.1%, notwithstanding the phase difference existing between the primary and secondary voltages of a GPT, is described. The method was used in field experiments with favorable results. The method is especially superior in terms of safety and operability because no contact with high-voltage portions is required     

The intravascular contrast agent Clariscan™ (NC 100150 injection) for 3D MR coronary angiography in patients with coronary artery disease

4. Conclusion In our experience Clariscan^TM improves image quality, thus, increasing the visual length and improving the actual visible dimensions of the coronary arteries. Distal segments improved to a higher extent than the proximal parts. In that way sensitivity and specificity for the detection of significant coronary artery disease could be improved. Better SNR and CNR can be used to improve spatial resolution or to reduce scan time by techniques like SENSE or SMASH [14,15].     

Magnetic resonance imaging of coronary arteries and coronary stenosis

Coronary artery imaging is an important investigation for the management of coronary artery disease. Alternative noninvasive imaging would be useful, but the small caliber and tortuosity of the coronary vessels and cardiac and respiratory motion create formidable imaging problems. We first studied 21 normal subjects and 5 with coronary artery disease established by X-ray contrast angiography, of whom 2 had undergone bypass grafting. Of these, 22 were imaged successfully. Identification of the artery was possible for the left main stem, left anterior descending, right coronary, and left circumflex arteries respectively in 95%, 91%, 95%, and 76%. The arterial diameter at the origin could be measured in 77%, 77%, 81%, and 63%. The mean ±SD arterial diameter in each case (4.8±0.8, 3.7±0.5, 3.9±0.9, and 2.9±0.6 mm) was not significantly different from reference values (allp=ns). The mean length of artery visualized was 10.4±5.2,46.7±22.8,53.7±27.9, and 26.3±17.5 mm. In 12 normal males, the total coronary area was 30.9±9.2 mm² and the ratio compared with body surface area was 16.4±4.4 mm² m^–2 (bothp=ns compared with reference values). In seven patients, with X-ray contrast coronary angiography, the proximal arterial diameter measured by magnetic resonance was 3.9±1.1 mm, and by X-ray contrast angiography 3.7±1.0 mm (p=ns). We then studied 17 patients with angina. Imaging of just the relevant artery was performed and analysis was blinded to the X-ray angiography results. Stenosis was identified on the magnetic resonance (MR) images by localized reduction in vessel signal intensity. Stenosis location by MR was assessed by measurement of its distance from a reference vessel, with correlation to the X-ray findings. X-ray coronary angiography showed 23 stenoses of which 15 (65%) were correctly located by blind assessment of the MR images. Of the eight remaining stenoses, a further 5 (63%) were correctly located on the MR images after retrospective comparison (overall sensitivity 87%). There were three lesions thought to represent stenosis by MR, which on review of the X-ray angiogram proved to be a minor stenosis <50% (two cases) or a tortuous vessel (one case). Greater signal loss was seen in the more severe stenoses. The stenosis length by MRI was greater than by X-ray (8.4 versus 5.1 mm,p<0.001). The overestimation of stenosis length may be due to turbulence.     

改进的基于特征融合MOSSE冠脉目标追踪算法

CT血管造影(computed tomography angiography,CTA)作为一种无创、检测精确较高的辅助诊断方法,尚急需能有效消除冠脉目标附近干扰噪声并寻求可全自动快速准确追踪目标的新算法以大幅减轻医生阅片压力、辅助其进行可靠诊断与治疗.提出了一种特征融合的误差最小平方和(minimum output sum of squared error,MOSSE)冠脉目标追踪新算法,通过提取冠脉血管多个特征,将其融合加入现有的MOSSE追踪方法,实现全自动准确快速追踪冠脉目标.使用河北大学附属医院9位患者(5男4女,均龄65岁,其中6位有冠心病史)的CTA数据进行了算法验证,并与文献已报道基于中心线提取和基于区域生长的现有冠脉目标提取算法进行了处理结果对比分析.结果表明,新算法处理追踪一例患者切片数据仅需耗时0.02 s,多个病例的平均准确度达94.30％,性能优于上述现有冠脉目标提取算法,能实现全自动准确高效追踪到形态变化剧烈的冠脉目标,可为冠心病的临床诊治起到更为高效的辅助作用.     

基于ST7MC的一种改进型反向电动势过零检测

介绍了基于ST7MC的反向电动势过零检测法,针对无刷直流电动机在低速和低电压应用情况下原方案的不足,提出了一种采用补偿放大电路的改进型反向电动势过零检测方法.