基于数据驱动的控制系统故障诊断研究

发布时间：2018-06-12 05:35

本文选题：控制系统 + 故障诊断　；参考：《华北电力大学》2017年硕士论文

【摘要】：随着科学技术的不断推进,现代控制系统的规模逐渐大型化,复杂程度也日益增大。为了提高控制系统的安全性和可靠性,需要对整个系统的运行状态进行监控,及时发现系统的故障信息,进而采取相应的措施,防止灾难性事故的发生。复杂的控制系统很难得到精确的数学模型,这导致传统的基于模型的故障诊断方法在应用上存在很大的局限性,而由于计算机技术的迅速发展,使得控制系统能够获得越来越多的数据信息,基于数据驱动的故障诊断方法就是在这种情况下应运而生的。本文以控制系统为研究对象,利用了基于数据驱动的方法进行故障诊断研究。首先,需要对控制系统进行故障检测,判断有无故障的发生。针对这一问题,提出了基于特征提取与聚类相结合的故障检测方法。该方法利用慢特征分析对控制系统的正常数据与故障数据进行特征提取,再采用模糊C-均值聚类算法,获得正常数据的聚类中心,并选取簇类半径作为阈值,构建待检测数据与聚类中心的差值模型,判断控制系统的待检测数据相对于正常状态是否发生了偏离。其次,构建控制系统故障库,对故障类型进行分类。整个系统由于引入了闭环控制规律,使得故障在系统内传播,最终可能使不同的故障产生相近的数据,因此将上述提取的故障特征与控制系统的故障特性相结合,提高故障特征的区分性。利用减法聚类的自适应特性,确定待检测的故障是否为故障库中已知的故障类型。然后再对待检测的故障数据进行模式匹配,判断故障类型。最后,将上述方法应用到电厂的过热汽温控制系统中进行方法验证。
[Abstract]:With the continuous advancement of science and technology, the scale of modern control system is becoming larger and more complex. In order to improve the security and reliability of the control system, it is necessary to monitor the running state of the whole system, find out the fault information in time, and then take corresponding measures to prevent the occurrence of catastrophic accidents. The precise mathematical model is rare in the hybrid control system, which leads to the limitation of the traditional model based fault diagnosis method, and because of the rapid development of computer technology, the control system can obtain more and more data information. The method based on the data driven fault diagnosis is in this case. This paper takes the control system as the research object and uses the data driven method to study the fault diagnosis. First, it needs to detect the fault in the control system and judge whether there is a fault. The characteristics of the normal data and the fault data of the control system are extracted, and then the fuzzy C- mean clustering algorithm is used to obtain the clustering center of the normal data, and the cluster radius is selected as the threshold, and the difference model of the data to be detected and the cluster center is constructed to determine whether the data to be detected in the control system occurs relative to the normal state. Secondly, the fault Bank of the control system is constructed, and the fault types are classified. The whole system is introduced the closed loop control law, which makes the fault spread in the system, and may eventually cause the different faults to produce the similar data. Therefore, the fault characteristics mentioned above are combined with the fault characteristics of the control system to improve the fault characteristics. By using the adaptive characteristic of subtraction clustering, it determines whether the fault to be detected is a known fault type in the fault bank. Then, the fault data is matched by mode and the type of fault is judged. Finally, the above method is applied to the superheated steam temperature control system of the power plant to verify the method.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP273;TP277

【参考文献】