基于分层策略与模糊五级SDG的故障定位方法

发布时间：2018-07-13 16:41

【摘要】：随着科学技术的不断进步，化工过程的复杂性不断提升，向着大规模、自动化的趋势发展。化工生产的原料易燃易爆，生产环境一般处于极端条件，因此故障会造成很大的危害。如何快速高效地推理并确定故障源已经成为了化工安全领域的研究热点。符号有向图(SDG)由于具有良好的信息完备性，不需要精确的数学模型即可快速地描绘出故障的传播过程，以及反映不同工作状态下变量间的相互作用关系等优点，被广泛应用于工业的故障诊断与故障定位。传统SDG包含的定性信息无法描述故障的严重程度，在进行复杂系统推理时需要大量时间，对于多个候选故障源无法进一步分辨。现有的SDG诊断方法大多从某一角度对传统SDG进行改进，会导致一些问题的出现，如对SDG模型进行信息扩充后，系统的诊断效率降低；忽略了故障动态信息，无法精确的定位故障源等。为了解决这些问题，本文提出了一种综合故障诊断方法，从模型、结构、推理三个方面对SDG模型进行了改进。传统SDG模型中，模型定量信息不足，故障的分辨率较差。变量节点只有三种状态，包含的故障信息较少。节点状态是根据阈值建立的，而阈值的设定是判断系统是否故障的重要标准。传统SDG模型的建立依托了单一阈值的设定，节点只描述了故障是否发生，，并没有反映出故障的严重程度。同时由于测量过程中噪声的存在，清晰值无法准确的描述故障信息，这就需要用更为完备的方法来描述系统变量。为了弥补SDG模型分辨率低及定量信息不足的缺点，本文将三级状态扩展为五级状态，同时引入了模糊集合理论，得到了一种模糊五级SDG模型。扩充模型信息会增大系统结构的复杂性，提高了搜索的难度。将SDG模型按照分层策略进行层次化处理，能够简化模型的结构。另外，分层后的SDG模型中，故障只能由高层向低层传播，从而能够快速得到候选故障源，减少了搜索空间。故障的搜索通常采用回溯搜索，但由于诊断信息不足，无法进一步搜索故障源。本文从推理角度引入了故障定位算法，通过公共点的设定，将故障源定位问题转化为相容通路优先级的排序问题，同时引入了故障放大倍数、故障支持度和故障速度来定位故障源，提高了故障定位的精度，弥补了传统SDG只包含静态信息的缺点。最后以TEP过程为例进行故障模拟分析，验证该方法的有效性。仿真结果表明，该方法高效准确，具有良好的信息完备性与故障定位能力，对减少设备损耗及提高化工系统的安全性起到了推动作用。
[Abstract]:With the development of science and technology, the complexity of chemical process is increasing and developing towards the trend of large scale and automation. The raw materials of chemical production are flammable and explosive, and the production environment is generally in extreme conditions, so the failure will cause great harm. How to quickly and efficiently infer and determine the fault source has become a research hotspot in the field of chemical safety. The symbol directed graph (SDG) has the advantages of good completeness of information, it can describe the process of fault propagation quickly without the need of accurate mathematical model, and reflect the interaction relationship between variables under different working conditions, and so on. It is widely used in industrial fault diagnosis and fault location. The qualitative information contained in the traditional SDG can not describe the severity of the fault. It takes a lot of time to infer the complex system and can not be further distinguished for multiple candidate fault sources. Most of the existing SDG diagnosis methods improve the traditional SDG from a certain point of view, which will lead to some problems. For example, after extending the SDG model, the diagnosis efficiency of the system is reduced, and the fault dynamic information is ignored. Unable to pinpoint the source of the fault, etc. In order to solve these problems, a comprehensive fault diagnosis method is proposed in this paper. The SDG model is improved from three aspects: model, structure and reasoning. In the traditional SDG model, the quantitative information is insufficient and the fault resolution is poor. Variable nodes have only three states and contain less fault information. The node state is established according to the threshold, and the threshold setting is an important criterion to judge whether the system is fault or not. The traditional SDG model is based on the setting of a single threshold, and the node only describes whether the fault occurs or not, and does not reflect the severity of the fault. At the same time, due to the presence of noise in the measurement process, the clear value can not accurately describe the fault information, which requires a more complete method to describe the system variables. In order to make up for the shortcomings of low resolution and insufficient quantitative information of SDG model, this paper extends the three-level state to five-level state, introduces fuzzy set theory, and obtains a fuzzy five-stage SDG model. Extending the model information will increase the complexity of the system structure and increase the difficulty of searching. The structure of SDG model can be simplified by hierarchical processing according to stratification strategy. In addition, in the layered SDG model, the fault can only propagate from the high level to the lower level, thus the candidate fault source can be obtained quickly and the search space can be reduced. Backtracking is usually used in fault search, but it is impossible to search for fault source because of insufficient diagnosis information. In this paper, a fault location algorithm is introduced from the point of view of reasoning. By setting common points, the problem of fault source location is transformed into a priority ranking problem of compatible paths, and the fault magnification is also introduced. Fault support and fault speed are used to locate the fault source, which improves the accuracy of fault location and makes up for the shortcoming that the traditional SDG only contains static information. Finally, the TEP process is taken as an example for fault simulation analysis to verify the effectiveness of the method. The simulation results show that the method is efficient and accurate, has good information completeness and fault location ability, and plays an important role in reducing equipment loss and improving the safety of chemical system.
【学位授予单位】：太原理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TQ086

【参考文献】