Design and experimental research of internal leakage detection device of buried pipeline ball valve based on valve cavity pressure detection

Valve is a key control component in the oil and gas long-distance pipelines, and the internal leakage detection of the valve has important research significance and application value. In this paper, a new method for detecting internal leakage of the buried pipeline ball valve based on the pressure change of the valve cavity was proposed. Firstly, a simplified internal leakage model of ball valve was established, and the change law of valve cavity pressure under different openings was obtained by finite element simulation. Secondly, a set of detection device was designed based on the internal leakage principle of buried ball valve and an experimental platform of ball valve internal leakage was built. Then, the pressure signal of the ball valve cavity under different openings and different pressures was collected by the pressure sensor, and the relationship between the power source pressure, the opening and the time required for the valve cavity pressure to reach equilibrium was established. Finally, the internal leakage degree of ball valve was simulated by ball valve opening and the Back Propagation (BP) neural network was selected to invert the ball valve opening. The inversion results showed that the maximum relative error of the liquid and gas experiment were 13.4% and 19.1%, respectively. The inversion results also verified the feasibility of BP neural network in the inversion calculation of the internal leakage degree of the ball valve, which provides guidance for the study of the classification of the internal leakage degree. The experiment and inversion results show that the proposed method can realize the detection of the internal leakage of the buried ball valve, which provides application reference in practice.     

Prediction method of ball valve internal leakage rate based on acoustic emission technology

Valves on natural gas transmission pipelines are prone to internal leakage due to the influence of environment, transportation medium, misoperation and other factors. Therefore, timely and accurate detection of valve internal leakage is of great significance to the safe operation of the pipelines. As a non-destructive testing method, acoustic emission (AE) technology can be used for online detection of internal leakage of valve. In order to improve the accuracy of AE technology for detecting valve internal leakage, a ball valve internal leakage detection method based on acoustic emission technology is proposed. Firstly, the AE signal generated by the internal leakage of the ball valve is detected by AE sensor, and the AE signal is denoised by wavelet packet threshold denoising. Secondly, the characteristic parameters with high correlation with the internal leakage rate of the ball valve are selected by using the correlation coefficient method to form the sample feature set. Finally, the mathematical model between the internal leakage rate of the ball valve and the characteristic parameters is established by using the sparrow search algorithm to optimize the backpropagation neural network, and to predict the internal leakage rate of the ball valve. An experimental platform is built to verify that the proposed method can predict the internal leakage rate of the ball valve more accurately. It provides strong technical support for the safe operation of natural gas transmission pipelines.     

神经网络在航空活塞发动机排气门故障预测中的应用

针对航空活塞发动机排气门卡阻故障,经过对故障机理的分析,提出了一种利用神经网络对排气门导套与气门杆的配合间障进行预测,以间接预测排气门卡阻故障的方法。将影响排气门积垢速率的因素设定合理的特征值,以这些特征值和发动工作时间作为输入向量,配合间隙作为输出向量,分别建立了GRNN神经网络和BP神经网络预测模型。预测实例表明,GRNN神经网络预测模型具有较高的预测精度、稳定的网络以及较快的收敛速度,预测性能优于BP神经网络模型,预测结果可作为评估排气门卡阻故障发生概率的有效依据。     

轧机伺服液压缸内泄漏故障诊断研究

针对目前轧机伺服液压缸故障诊断过程中,故障特征提取困难,信号非线性变化,数据量大的问题,提出了一种基于深度置信网络的轧机伺服液压缸故障诊断的方法。根据轧机系统工作原理,建立轧机系统仿真模型,对轧机内泄漏故障状况进行模拟。利用深度置信网络在智能故障诊断的优越性,将信号归一化处理后放入深度置信网络进行训练,然后通过反向传播学习,优化网络各参数,提高诊断精度。深度置信网络模型由多层玻尔兹曼机以及顶层BP神经网络组成。与传统BP神经网络方法进行比较,结果表明,在训练样本数据足够的条件下,深度置信网络模型在伺服液压缸内泄漏故障诊断具有更高的诊断精度。     

基于BP神经网络的液压缸内泄漏诊断

液压缸是工程机械中常用的执行元件,内泄漏是其常见的故障模式,将严重影响机械系统的工作效率和安全性,及时识别液压缸内泄漏能够保证液压缸的安全正常工作。通过小波分解提取液压缸进口压力信号特征,利用BP神经网络建立分类器,实现了对液压缸内泄漏的智能识别,分类准确率高,提高了液压缸内泄漏故障诊断的效率,为实现液压缸智能状态监控提供了基础。     

基于神经网络的液压缸内泄漏在线测量研究

内泄漏是液压缸最常见故障，严重影响液压缸的正常工作，因此对其在线测量显得尤为重要。提出内泄漏在线测量的工作原理，包括在线测量系统、应变片的固定方式和流量-应变信号转换的数学模型，并搭建实验系统采集内泄漏和应变数据并进行数据处理。分别采用BP神经网络和卷积神经网络对液压缸内泄漏进行预测，结果表明，卷积神经网络准确度高、效率高，为其他液压元件微小流量的在线测量提供一种新的思路。     

基于EEMD_BP网络的滚珠丝杠副故障模式识别

针对在机械故障诊断领域,对信号的时频域处理分析提取特征值往往不能准确判断机械故障状态的问题。在对数控机床滚珠丝杠副振动信号研究中,提出了利用集合经验模态(EEMD)方法分析受到噪声干扰的3种不同状态的滚珠丝杠副振动信号。利用BP神经网络理论,以振动信号的时频域特征值及EEMD分解得到内禀模态函数(IFM)特征值作为输入,建立BP神经网络模型,并通过实验验证诊断网络模型的可靠性。     

基于蚁群神经网络的滚动轴承故障诊断

为了提高滚动轴承故障诊断的准确性,将蚁群算法与神经网络相结合,根据轴承故障产生的机理,建立其BP神经网络的诊断模型,以网络的误差为目标函数,通过蚁群算法进行BP网络的权值优化,并用优化好的BP网络进行故障诊断。仿真结果表明,该方法具有较高的故障诊断准确度,具有较强的实用性。     

基于BP神经网络的管道泄漏声信号识别方法研究

针对城市供水管网泄漏检测需求,进行了泄漏声信号识别方法研究。分析了泄漏信号的时域、频域及波形特点,提取出可用于泄漏信号表征的20种特征参数;基于提取的泄漏声信号特征参数,构建了泄漏声信号BP神经网络识别系统;研究了神经网络结构(隐含节点数、传递函数、学习率)及输入参数的数量和种类对泄漏信号识别效果的影响,并优化出最佳的神经网络结构及输入参数。在以上研究基础上,利用优化后的神经网络对实验室及现场管道泄漏信号进行了交叉训练和识别,结果表明,提出的基于泄漏特征参数的神经网络系统具有较高的可靠性和普适性,可以很好地实现不同场景下泄漏信号的交叉识别,整体识别率达92.5%,为解决不同工况下泄漏信号识别做了有益的探索。     

Study on the design of ball valve based on elastic ring valve seat structure and fluid characteristics and fatigue strength

Aimed at the technical problems such as the influence of granular medium on spring pre-tightening force sealing, a new ball valve based on elastic ring valve seat structure is studied. The spring plate type valve seat structure is designed to cooperate with the ball core for sealing, and the blade spring coil is used to cooperate with the ball core for sealing in the spring plate type valve seat structure. Wherein the supporting back ring supports the blade leaf spring on the outer side to enhance and protect the role of the blade spring coil. The design without the spring cavity avoids the problem of sealing failure caused by medium entering into the spring cavity and affecting the compression spring, and avoids the situation that the valve seat can be sealed with the ball core by pre-tightening the compression spring, thus avoiding the problem of sealing failure caused by the valve seat sticking on the valve body. The mechanical and flow characteristics are studied and analyzed by the ball valve characteristic test system. The stem torque, unbalance torque, flow characteristics and flow coefficient variation at different nominal diameters are analyzed. The seal allowable squeeze stress and seal surface pressure are analyzed, and the seal is stable and reliable with the seal pressure meeting the seal design criteria. The fluid dynamics simulation analyzes the velocity, pressure and flow traces of the fluid flowing through the ball valve under three opening degrees: fully closed, half open and fully open, the maximum velocity-pressure and opening degree variation curves of the inlet and outlet, the maximum velocity-pressure and opening degree variation curves of the inlet and outlet under different nominal diameters and the flow resistance coefficient curves. Static strength analysis was done for the ball core and spring plate seat structure to obtain the stress, displacement, strain and safety factor. The fatigue strength of the ball spool and spring-loaded plate seat structure was analyzed, and the total number of lives (cycles) and load factors were obtained, and the results show that the fatigue strength of the ball spool and spring-loaded plate seat structure is safe and the fatigue strength meets the requirements. Ball valve pressure test, low pressure sealing test and high pressure sealing test, valve body strength and ball valve sealing performance all meet the requirements.     

BP神经网络在截齿合金头失效识别中的应用

为实现采煤机截割过程中截齿合金头失效形式的监测和识别,提出一种基于BP神经网络的多特征信号识别截齿合金头失效形式的方法。测试提取截割过程中合金头龟裂、合金头脱落、合金头崩刃和合金头严重磨损4种截齿x,y,z三个方向上的振动特征信号和截割电机电流特征信号,选取特征值信号的最大值、均值和方差作为特征样本对BP神经网络进行学习和训练,建立截齿合金头失效形式的识别模型,实现截割过程中截齿合金头失效形式在线监测与准确识别。实验结果表明,BP神经网络的判别结果和测试样本的实际失效类型相符,能够对截齿合金头失效形式进行准确识别,为实现采煤机截齿在线监测和失效形式识别提供新的方法和手段。     

储罐底板漏磁检测量化技术研究

储罐是石油、石化工业中重要的设备,储罐底板腐蚀是储罐安全隐患之一。漏磁检测方法是目前储罐底板检测研究的一个重要方向。根据缺陷漏磁信号的特征,将经验模态分解方法（EMD）与小波去噪方法相结合,对漏磁信号进行去噪处理。采用BP神经网络模型对储罐底板缺陷进行量化分析研究,构建了缺陷几何参数预测BP神经网络模型,并运用有限元分析所得到的数据为BP网络训练样本,用人工模拟缺陷的漏磁信号测试BP神经网络。网络训练和测试结果符合储罐底板缺陷量化的精度要求。     

基于神经网络与PSO算法的发动机装配工艺参数优化

针对汽车发动机装配过程中缸体泄漏问题,结合Back Propagation(BP)神经网络及粒子群优化(Particle Swarm Op-timization,PSO)算法,提出了一种发动机装配工艺参数优化方法.首先,使用BP神经网络建立了生产工艺参数与质量指标之间的非线性映射关系,并以此作为泄漏率预测模型.其次,根...     

多几何要素影响下液压阀件特性的混合神经网络预测模型

液压阀件系统是一个具有多几何要素影响的多系统特性复杂系统,建立液压阀件特性预测模型,以系统多几何要素作为输入,实现系统特性的预测,将对实际生产具有重要的意义。在深入分析反向传播(Back propagation,BP)神经网络与径向基函数(Radial basis function,RBF)神经网络的基础上,结合两类神经网络的特点,提出基于BP神经网络与RBF神经网络的混合神经网络预测模型。利用生产实际中实测的某具体液压阀件特性值及影响该特性的各几何要素值作为预测模型的数据来源,对所提出的混合神经网络进行训练,并进行仿真。实例计算表明混合神经网络预测模型可提高单项神经网络预测模型的预测精度,预测平均相对误差为0.0461。可见,所提出的混合神经网络预测模型能够很好地满足工程实践中液压阀件特性预测要求。     

双隔离管线球阀腔体泄放压力的规定与分析

介绍了双隔离管线球阀腔体泄放压力在相关标准中的规定。提出了阀门泄放压力超过失火状态规定的额定压力许用值及存在的潜在危险。分析了双隔离管线球阀在非正常工况下压力边界可能失效的原因。     

基于BP神经网络安全阀的失效评价

在实际使用过程中,安全阀的失效涉及诸多因素,且易受随机性因素的影响,具有模糊性和不确定性的特点,这些特点决定了整个系统是非线性动力学的.将安全阀失效的相关影响因素作为因素集,应用BP人工神经网络模型,利用MATLAB神经网络工具箱GUI,对样本数据进行仿真,然后采用训练好的网络对安全阀的现状进行评价.BP神经网络强大的记忆功能使得能从安全阀以往失效情况中总结出一般的失效规律,快速、准确地对安全阀现状做出判断.将BP神经网络用于安全阀失效评价,具有良好的评价效果和重要的实用价值.     

基于ReCorre特征优化的气阀故障诊断方法

振动信号包含信息丰富,反应状态直接,振动法是压缩机气阀故障诊断常用方法.振动信号的特征参数繁多,特征向量选择是否合理对故障诊断结果准确性影响很大.研究ReCorre方法对气阀振动信号特征参数进行优化选择,再通过神经网络进行分类识别.实例表明,基于特征优化的模糊神经网络分类识别结果正确率高,识别结果受数据来源影响小,是一种较好的气阀故障诊断方法.     

基于小波变换预处理的一种神经网络故障诊断方法的研究

针对无法区分干扰信号与有用信号一类系统提出了一种先用小波变换让信号在不同频段上进行分解,提出各分量故障信号的特征,然后应用ＢＰ神经网络进行故障诊断的一种方法。并将这种方法应用于液压系统的泄漏故障诊断,取得了比较满意的效果。     

盾构机液压推进系统故障分析与仿真研究

针对盾构机液压推进系统的故障诊断问题,以在芜湖长江隧道项目过程中所使用的大型液压驱动型泥水盾构机为研究对象,分析了盾构机液压推进系统的工作原理,总结了推进系统中液压缸泄漏、换向阀泄漏及溢流阀泄漏故障模式的发生机理及其对推进系统造成的影响。利用AMESim平台建立推进系统模型,对液压缸泄漏、换向阀泄漏及溢流阀泄漏3类故障进行仿真分析,并提取液压缸推进速度、推进行程、无杆腔流量和系统压力4种推进参数的仿真数据。仿真结果表明:发生液压缸泄漏故障时,活塞杆无法伸出,推进速度为0;发生换向阀泄漏故障时,液压缸出现自走现象,推进速度明显降低;发生溢流阀泄漏故障时,系统压力明显降低,液压缸无法克服阻力向前推进。为后续盾构机推进系统的故障诊断和预测提供了有价值的参考。     

基于小波包BP_AdaBoost算法的机载燃油泵故障诊断研究

针对机载燃油泵故障数据少、诊断效率低、维护成本高、缺乏有效诊断方法的问题,搭建了机载燃油泵燃油转输系统实验平台,提出利用小波包分析进行特征提取和基于BP_AdaBoost机载燃油泵故障诊断方法。首先测量燃油泵7种典型状态模式所对应的振动信号和出口压力信号;然后在分析信号时频特性和统计特性的基础上,利用小波包分解提取振动信号不同频段能量值作为故障特征参数,结合振动信号峭度以及压力信号均值构造特征向量;最后利用特征向量训练和验证BP_AdaBoost分类模型。实验结果不仅优化了传感器,而且表明BP_Adaboost算法与SVM、BP算法相比,能够有效实现对机载燃油泵的故障诊断。