Study on Feasibility of Hybrid Fault Diagnosis system

1.混合故障诊断系统的设计(1)建立了混合故障诊断系统的基本设计方法。基本战略包括两个层面。上层由基于模型的故障诊断算法组成。底层由基于知识的故障诊断算法和基于数据的故障诊断算法组成。(2)将基于符号有向图的故障诊断算法作为上层策略，提出了一种估计混合故障诊断系统详细设计所需的最大候选集的方法。(3)给出了(2)中提出的方法在乙酸法的氧化子过程中的实际应用实例。混合故障诊断系统的性能分析(1)在PC机上实现了混合故障诊断系统。对醋酸过程氧化子过程中的各种故障进行了仿真，以评估所提出的混合系统的速度和诊断准确率。(2)通过对发生堵塞和泄漏的油罐-管道系统的实验，对所提出的混合系统的速度和诊断精度进行了评估。(3)在速度方面，在模拟和实验的情况下，使用个人计算机在大约120秒内完成诊断。在精度方面，仿真和实验中考虑的每个真实原因都得到了正确的识别。结果表明，所提出的混合故障诊断系统可以应用于实际对象。

1. Design of hybrid fault diagnosis system (1) A method of basic design of hybrid fault diagnosis system has been established. The basic strategy comprises two-levels. The upper-level consists of model-based fault diagnosis algorithm. The lower-level consists of knowledge-based or data-based fault diagnosis algorithm. (2) A method to estimate the greatest sets of candidates, which should be prepared for detailed design of hybrid fault diagnosis system, has been developed for using the fault diagnosis algorithm based on the signed directed graph as the upper-level strategy. (3) An example of the practical application of the method developed in (2) was presented for an oxidation subprocess of an acetic acid process.2. Performance analysis of hybrid fault diagnosis system (1) The hybrid fault diagnosis system was implemented on the personal computer. Simulations were performed for a various types of faults caused in an oxidation subprocess of an acetic acid process to evaluate the speed and the diagnostic accuracy of the proposed hybrid system. (2) Also, the speed and the diagnostic accuracy of the proposed hybrid system were evaluated by experiments u sing a tank-pipeline system, where blockages and leakages were generated. (3) As to the speed, diagnoses were completed within about 120 seconds in both cases of simulations and experiments by using a personal computer. As to the accuracy, every real cause considered in the simulations and experiments were identified correctly. As a result, the proposed hybrid fault diagnosis system can be applied to practical plants.