The Exploration Of Intelligent Fault Diagnosis Based On Outlier

With rapid development of modern science and technology and increasing complexity of engineering equipments, the modern process industry continues to expand its scale and become increasingly complex and integrated, the degree of automation is higher and the safety and reliability of equipment operation is also drawing greater attention. Meanwhile, due to structural complexity of equipments and continuous high-power and high load operation, malfunctions will inevitably occur during operation with accumulated working hours and internal and external conditional changes of the equipments. Once such systems fail, the minimum consequence would be lowered equipment performance which affects production while the maximum impact would be halt or cessation of production, equipment damage or even death caused by equipment failures. Such malfunctions will not only result in huge stuff and property losses, but also cause irreversible damage to the ecological environment. In the past 30 years, traditional fault diagnosis technology has experienced rapid development and played a tremendous role in various projects. Traditional techniques of fault diagnosis can play their unique role when dealing with the relatively simple equipments or a single set, but they are powerless to treat the large-sized sophisticated equipments or the cases of multi-malfunctioning, especially so in the absence of specialized fault detection equipments. In these cases, using intelligent fault detection techniques based on abnormal data can analyze the existing data only and extract the abnormal data. By doing so and then conduct the intelligent examination and analysis can simplify fault diagnosis and obtain greater exactness in fault treatment.In this paper, the author first gives an overview of the definition of the intelligent fault and some methods of conducting the intelligent fault diagnosis, illustrates the significance of such kind of diagnosis, analyzes the advantages and disadvantages of the existing fault diagnosis systems and puts forward the method of using abnormal data as samples for intelligent fault diagnosis so as to realize the efficiency and simplification of diagnosis and the minimum occupation of equipments; second, the author conducts Anti-Jamming to treat data traffic so as to strengthen the validity of abnormal data and extract clean and reliable data as measurements; the author then elaborates on the definition of abnormal data and the method of inspecting them, which is using partial least-squares as the inspecting method; and then, using the abnormal data extracted in this way as samples, the author conducts an intelligent fault diagnosis by ANFIS and sites one examples to make comparison. Finally, the author points out the relevance among faults and uses fuzzy association rules to infer malfunctions that has been overlooked or those emerging, so as to enhance the accuracy of fault diagnosis and the effectiveness of fault prediction.