机械密封端面状态监测及寿命预测关键技术研究

发布时间：2018-01-18 03:12

本文关键词：机械密封端面状态监测及寿命预测关键技术研究　出处：《西南交通大学》2015年博士论文　论文类型：学位论文

【摘要】：机械密封(Mechanical Seal,MS)是反应釜、离心泵、压缩机和涡轮机等机械装置的组成部件,用于旋转设备防止泄露、节约能源和保护环境等。如果把反应釜、离心泵、压缩机和涡轮机等看做一个动力系统的心脏,那么机械密封就相当于这颗心脏的瓣膜。机械密封状态监测(Mechanical Seal Condition Monitoring,MSCM)的目的在于掌握密封设备发生故障之前的异常信息与劣化征兆,以便工程人员事前采取针对性措施防止故障的发生或降低其发生的可能性,从而减少因故障所导致的停机时间和因停机而造成的所有损失,最终降低企业维修费用并提高设备的有效利用率。本文在深入探讨国内外机械密封状态监测研究现状的基础上,通过多传感器监测实验对机械密封端面状态进行了进一步的研究分析,其主要内容如下：(1)机械密封端面一方面要接触紧密以防止流体泄露；而另一方面密封端面间隙又不能过小,以防止端面过度接触而损伤密封端面。为了保障机械密封低摩擦和小泄露量的兼容和平衡,首先须对设备运转时密封端面开启转速进行检测。由于声发射信号对机械密封端面接触状态具有很好的直接反馈性,而且双边Laplace小波能够很好地提取声发射信号中的脉冲性,提出基于声发射信号的双边非对称Laplace小波相关系数法用来检测机械密封端面开启转速。(2)机械密封端面接触状态直接影响着密封装置性能和使用寿命,因而对机械密封端面摩擦状态的监测非常必要。由于声发射信号频谱受主轴转速影响非常小,研究了基于声发射信号利用超球面支持向量机(Hyper-Sphere Support Vector Machine, HSSVM)对密封端面接触状态的识别方法。试验中以低转速、非接触摩擦状态声发射信号小波包能量特征训练超球面支持向量机,能够识别不同转速下密封端面接触状态。(3)机械密封端面磨损程度不仅影响密封工作状态,而且决定密封泄漏量的大小。为了在没有全历程状态数据的情况下对机械密封端面做出较为准确地磨损状态评估,提出了基于偏置因子隐马尔可夫模型(Bias Factorial Hidden MarkovModel,BFHMM)的机械密封磨损状态评估技术。传统的基于完备先验知识的监测模型建立非常困难,而且普适性较差,本文研究在仅具备机械密封端面最初和严重磨损两种状态下的先验知识,即在没有全历程状态数据的情况下对机械密封运行状态作出较为准确地评估,从而避免机械密封过度磨损,保证密封的正常使用和设备的安全运行。(4)对机械密封的使用剩余寿命的预测是评定一套机械密封性能的重要指标。考虑到灰色理论能有效预测信号的总体趋势,而粒子滤波却对其局部变化更为敏感。本文将两种方法相结合,提出灰色粒子滤波预测方法：首先利用粒子滤波对曲线拐点进行跟踪修正,然后利用灰色理论拟合预测密封摩擦副的总体寿命曲线。数据证明,该方法能对机械密封摩擦副的剩余寿命进行有效预测。本论文研究内容主要完成了对机械密封端面开启转速检测,建立了密封端面接触状态和磨损状态的识别评估模型,实现了对端面接触状态监测、端面磨损程度评估及摩擦副剩余寿命预测,为以后工业现场机械密封状态监测实施提供了理论依据和技术支持。
[Abstract]:Mechanical seal (Mechanical Seal MS) is a reaction kettle, centrifugal pump, compressor and turbine components such as mechanical device, rotary device for preventing leakage, save energy and protect the environment. If the reaction kettle, centrifugal pump, compressor and turbine etc. as a dynamic system of the heart, so the mechanical seal is in the heart of the valve. The mechanical seal condition monitoring (Mechanical Seal Condition Monitoring, MSCM) is to control the abnormal information and signs of deterioration of sealing equipment failure before, so that the project staff take place or reduce the possibility of targeted measures to prevent failure, thereby reducing due to failure caused by downtime and all the losses caused by downtime, reduce maintenance costs and improve enterprise finally the effective utilization rate of the equipment. This paper deeply discusses the domestic and foreign machinery in sealed condition Based on the research status of monitoring, through multi sensor monitoring experiments were conducted to further study on end face mechanical seal, the main contents are as follows: (1) mechanical seal to a close contact to prevent fluid leakage; on the other hand, seal gap can not be too small, to prevent excessive contact face in order to protect the damage the seals. Compatible with mechanical seal of low friction and a small amount of leakage and balance, must first be on the seal face opening speed of equipment operation for testing. Due to its direct feedback is very good on the mechanical seal face contact state of the acoustic emission signals, and bilateral Laplace wavelet can be well extracted from acoustic emission pulse in the signal, proposed to the bilateral asymmetric Laplace wavelet correlation coefficient method of acoustic emission signal detection based on the mechanical seal face opening speed. (2) mechanical seal end face contact The state directly affects the sealing performance and service life, so it is necessary to monitor the mechanical seal friction state. Because the acoustic emission signals from the spindle speed effect is very small, is studied by using the hypersphere support vector machine based on acoustic emission signals (Hyper-Sphere Support Vector Machine, HSSVM) on the recognition method of seal face contact state. During the experiment, low speed, non contact friction state of AE signal characteristics of wavelet packet energy training hypersphere support vector machine, to seal the contact state recognition under different speed. (3) mechanical seal wear not only affect the sealing state, and seal leakage rate. In the absence of the whole process to state data under the condition of mechanical seal wear to make a more accurate assessment, the bias factor of hidden Markov model based on Bias (Fa Ctorial Hidden MarkovModel, BFHMM) of the mechanical seal wear condition assessment. The traditional monitoring model based on the complete prior knowledge is very difficult, and poor universality, in this paper, only with the mechanical seal and the initial state of the serious wear of two kinds of prior knowledge, the whole process of state data. The mechanical seal running state to make more accurate evaluation, so as to avoid excessive wear of the mechanical seal, to ensure the safe operation of normal use and equipment sealing. (4) the residual life prediction of mechanical seal is an important index to evaluate a set of mechanical sealing performance. Considering the grey theory can effectively predict the overall trend signal but, but the local particle filter is more sensitive to the change. This paper combines the two methods, the gray prediction method is proposed: firstly, particle filter particle filter on. Line tracking inflection point correction, and then use the grey theory prediction of overall life curve fitting the sealing friction pair. Data show that this method can effectively predict the residual life of mechanical seal friction pair. The main contents of this dissertation were completed on the mechanical seal face opening speed detection, establishes the identification evaluation model of the contact state and the seal face wear condition, the contact state monitoring of end face wear and friction degree evaluation of residual life prediction, provide a theoretical basis and technical support for future industrial field monitoring implementation of the mechanical seal.

【学位授予单位】：西南交通大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：TH136

【相似文献】