基于多元状态估计的电站风机故障预警研究及系统开发

发布时间：2018-07-25 12:32

【摘要】：风机是火电站重要的辅机设备,其运行状态直接影响电力生产的经济性和安全性。随着大容量、高参数火电机组的快速发展,电厂对风机设备的可靠性提出更高要求。当前,工业设备状态识别技术正由状态监测、故障诊断向故障预警发展。本文主要研究基于多元状态估计技术的电站风机故障预警方法,可以赢得宝贵时间采取措施降低故障损失或者避免故障发生,给发电企业带来巨大经济效益。研究风机结构和常见故障,分析故障所表现的监测信号,并且归纳当前电站风机可获得的监测信号。在此基础上,根据“可获得”、“故障敏感”、“最简”原则挑选风机MSET建模变量。然后对历史数据进行剔除异常数据、轴承温度3取1、标准化三项预处理,并且提出动态过程记忆矩阵构建法建立风机正常状态MSET模型。最后,以某电厂引风机正常状态历史数据进行建模仿真,验证所建立风机状态模型具有很高精度,可满足风机故障预警应用要求。MSET风机状态建模研究表明观测向量和估计向量之间的差异隐含丰富故障信息。为了充分挖掘故障信息和捕捉故障发展过程,提出关于观测向量和估计向量的相似度函数用于定量衡量二者差异,并且根据各变量对故障预警的重要程度,利用层次分析法确定相似度函数中各变量权重。然后,利用滑动窗口统计法降低随机干扰影响,根据正常状态平均相似度边界值确定合理故障预警阈值。从而提出基于MSET风机模型故障预警方法,如果新观测向量对应的平均相似度超越预警阈值,那么就可发出故障报警提醒运行人员处理。最后,以某电厂风机某次故障和三类模拟故障为例进行应用研究,结果验证所提出的方法能够发现风机早期故障,实现准确的风机故障预警。利用本文所提的预警方法和B/S结构开发一套电站风机故障预警系统,实际应用表明本文所提出的基于MSET风机状态模型的故障预警方法可以实现准确的风机故障预警,同时为其它工业设备故障预警提供一种可行的解决方案。
[Abstract]:Fan is an important auxiliary equipment in thermal power station. Its running state directly affects the economy and safety of electric power production. With the rapid development of large capacity and high-parameter thermal power units, the reliability of fan equipment is required higher in power plants. At present, the status recognition technology of industrial equipment is developing from condition monitoring and fault diagnosis to fault warning. This paper mainly studies the fault warning method of power plant fan based on multivariate state estimation technology, which can save valuable time to take measures to reduce the fault loss or avoid the fault, and bring huge economic benefits to the power generation enterprise. In this paper, the structure and common faults of fan are studied, the monitoring signal of fault is analyzed, and the monitoring signal that can be obtained by fan in power station is summarized. On this basis, according to the principles of "available", "fault sensitivity" and "minimization", the fan MSET modeling variables are selected. Then the historical data are eliminated, bearing temperature 3 is taken 1, standardized three items are preprocessed, and the dynamic process memory matrix construction method is put forward to establish the MSET model of fan normal state. Finally, using the historical data of the normal state of the induced fan in a power plant for modeling and simulation, it is verified that the established fan state model has a high accuracy. The research on the state modeling of MSET fan shows that the difference between the observation vector and the estimation vector is rich in fault information. In order to fully mine fault information and capture fault development process, the similarity function of observation vector and estimation vector is proposed to measure the difference between them quantitatively, and according to the importance of each variable to fault early warning. The weight of each variable in the similarity function is determined by analytic hierarchy process (AHP). Then the sliding window statistics method is used to reduce the influence of random interference and the reasonable threshold of fault warning is determined according to the average similarity boundary value of normal state. A fault warning method based on MSET fan model is proposed. If the average similarity of the new observation vector exceeds the warning threshold, then the fault alarm can be issued to the operator to deal with it. Finally, an application study is carried out on a fan fault and three kinds of simulated faults in a power plant. The results show that the proposed method can detect the early fault of fan and realize accurate early warning of fan fault. Using the early-warning method and B / S structure proposed in this paper, a set of power fan fault warning system is developed. The practical application shows that the fault early warning method based on MSET fan state model can realize accurate fan fault early warning. At the same time, it provides a feasible solution for other industrial equipment failure warning.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TM621

【相似文献】