基于人工神经网络的整流电路故障诊断

提出了一种基于人工神经网络进行整流电路故障诊断的方法.分析了变频器整流电路的故障模式,提出了根据直流母线电压频谱作为神经网络的输入进行故障定位的方法,并进行了仿真研究.     

Fault detection and isolation for a nonlinear railway vehicle suspension with a Hybrid Extended Kalman filter

Fault detection is considered to be one way to improve system reliability and dependability for railway vehicles. The secondary lateral and anti-yaw dampers are the most critical parts in railway suspension systems. So far, the dampers have been modelled as linear components in the fault detection and isolation observer design. In this work, a Hybrid Extended Kalman filter is used to capture the nonlinear characteristics of the dampers. In order to detect and isolate faults, a nonlinear residual generator is developed, which can distinguish clearly between different types of faults. A lateral half train model serves as an example for the proposed technique. The results show that failures in the nonlinear suspension system can be detected and isolated accurately.     

A real-time impact detection and diagnosis system of catenary using measured strains by fibre Bragg grating sensors

This paper describes an impact detection system using strain signals based on fibre optic sensors(FBG) for the real-time monitoring of the catenary system. The proposed detection system consists of three subsystems: a measuring system, a data processing and analysis system, and a status display and data access system. Because the strain signals obey the normal distribution, to monitor the catenary system in real time, a novel method that combines mobile standard deviation with the mobile Pauta criterion is proposed to distinguish real impact from the strain signal background. The use of this adaptive judging method reduces the misjudgment rate of impacts and improves the impact recognition accuracy. These impacts can be identified by the data analysis system, which provides impact location and their causes using the features of the catenary system. This method can simplify the detection system compared with the traditional location method. An application to a commercial metro line system indicated that the impacts on the catenary system were main caused by overlaps, expansion joints or steady arms, and were verified by correspondence with the floor plan of the catenary and manual inspection results. These results verified the reliability and effectiveness of the proposed impact detection system.     

浅谈市区沥青路面病害成因及处治措施

道路是城市的动脉,沥青路面因其噪音低、舒适性高的特点得以在市区快速推广。该文以惠州市的市区沥青路面的病害为例,总结了市区沥青路面病害的特点、成因及处置维修措施。对城市沥青路面的管养维护总结了较丰富的实践经验,以供同行借鉴。     

船用柴油机故障诊断的综合法研究

文章针对当前船舶柴油机常用故障诊断方法的基本原理，总结了各种方法的优缺点，提出了故障诊断综合法。运用该方法对两起故障进行了分析，结果表明综合法对及时发现并排除故障是有效的，具有一定的现实意义。     

接触网可靠性研究方法浅析

介绍了国内外接触网可靠性研究的现状,分析了接触网的主要故障、接触网强度可靠性分析的计算模型和可靠性计算方法,对接触网可靠性研究存在的难点及其进一步研究的方向提出了几点建议。     

基于组合式BP网络的空压机故障诊断

针对空压机故障诊断过程中参数的多样性及复杂性,分别采用单BP网络和组合式BP网络两种方法对其主要故障进行建模和诊断。结果表明:采用组合式BP网络的故障诊断方法更加简单实用,能有效地提高空压机故障诊断的准确率。     

基于HMM的客车转向架故障诊断与应用

介绍隐马尔可夫模型（HMM）算法在转向架故障诊断中的应用.利用HMM方法研究转向架在运行过程中抗蛇行减振器正常、异常状态,HMM的参数根据试验数据统计训练出,提出了一种新的转向架故障诊断方法,试验证明该诊断方法可行.     

一种基于MPLS-DiffServ流量工程的故障恢复模型

针对基于MPLS流量工程的本地、路径、重路由故障恢复模型存在资源利用率不高、恢复周期较长的缺点,提出了一种基于MPLS-DiffServ流量工程的故障恢复模型。该模型根据网络的故障发生概率、故障恢复时延的要求与网络资源利用率等参数和数据流分类进行配置,选择路径恢复算法。结果表明,该模型提高了资源利用率,缩短了故障恢复周期,对网络数据流故障恢复具有较强的有效性。     

日产Y50风雅汽车后排电动座椅的故障诊断

结合日产Y50风雅汽车后排电动座椅电路图，分析日产风雅汽车后排电动座椅工作原理与操作，详细阐述日产风雅汽车后排电动座椅出现故障时的检查步骤。