基于WFCM-GRA的机械产品故障诊断方法研究

发布时间：2018-03-11 09:20

本文选题：机械产品　切入点：故障诊断　出处：《合肥工业大学》2017年硕士论文　论文类型：学位论文

【摘要】：机械产品在生产生活中应用广泛,其运行状态直接影响着设备的正常运行及生产生活的质量。机械产品发生故障常会产生次生故障,对设备安全和生产生活的潜在危害大,因此,实现机械产品的实时状态监测与快速高效地故障诊断对保证设备安全连续运行,减少机械产品故障带来的潜在损失和威胁,降低其寿命周期内运行总成本,延长服役周期具有重要的现实意义。本文在过程监测的基础上,建立基于加权模糊C均值聚类(FCM)和综合关联分析的两阶段故障诊断模型,实现了机械产品的故障诊断。本文的主要内容包括:(1)针对不同样本和特征在过程监控和故障分类识别时的重要程度不同的问题,提出基于J指标的特征选取方法选取关键特征参数,采用Relief F方法和样本点密度的样本加权方法以突出关键特征参数和典型样本对分类识别的影响。(2)提出根据样本数据的不同特征参数统计特征,选取合适的联合控制图方案实现对机械产品运行过程的监控和预警,并在发动机缸体缓变故障监测过程中验证了本文提出的联合控制图方案的合理性。(3)针对机械产品故障诊断的特点,建立了两阶段的基于加权模糊C均值聚类-灰色关联分析(WFCM-GRA)的故障诊断模型:在基于WFCM和聚类有效函数获取各运行状态数据的聚类中心作为最具代表性的标准序列的基础上,运用综合灰色关联方法分析待检测样本数据与各标准序列的综合关联度,再根据综合关联序确定可能的故障类型,并在发动机故障诊断案例中验证了本文故障诊断方法的有效性和实用性。并在本文提出方法的基础上,开发了发动机过程监控预警和故障诊断的系统。
[Abstract]:Mechanical products are widely used in production and life, and their running state directly affects the normal operation of equipment and the quality of production and life. Therefore, the realization of real-time state monitoring and rapid and efficient fault diagnosis of mechanical products can ensure the safe and continuous operation of equipment, reduce the potential losses and threats caused by mechanical product failures, and reduce the total operating cost of mechanical products during their life cycle. It is of great practical significance to extend the service cycle. Based on the process monitoring, a two-stage fault diagnosis model based on weighted fuzzy C-means clustering (FCM) and comprehensive correlation analysis is established in this paper. Fault diagnosis of mechanical products is realized. The main contents of this paper include: (1) aiming at the problem that different samples and features have different importance in process monitoring and fault classification, A method of selecting key feature parameters based on J index is proposed. Relief F method and sample weighting method of sample point density are used to highlight the influence of key feature parameters and typical samples on classification recognition. Select appropriate joint control chart scheme to realize the monitoring and early warning of the mechanical product operation process. The rationality of the joint control chart scheme proposed in this paper is verified in the process of engine cylinder block slow change fault monitoring. (3) aiming at the characteristics of mechanical product fault diagnosis, A two-stage fault diagnosis model based on weighted fuzzy C-means clustering and grey correlation analysis (WFCM-GRA) is established. Based on WFCM and clustering effective function, the clustering center of each running state data is obtained as the most representative standard sequence. The synthetic grey correlation method is used to analyze the comprehensive correlation degree between the sample data to be tested and each standard sequence, and then the possible fault types are determined according to the comprehensive correlation order. The effectiveness and practicability of this method are verified in the case of engine fault diagnosis. Based on the proposed method, an engine process monitoring, warning and fault diagnosis system is developed.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TH17

【相似文献】