The Research Of Vibration Fault Diagnosis System For 900MW Turboset Based On Data Mining

Turbine as one of the core equipment of power plants long-term and high-speed work in poor conditions, rotor vibration appear higher the probability of failure, and vibration fault often pose a major hidden safety in production. So the vibration problem of turboset has always been the people of the important issues. The technology of the treatment, analysis and diagnosis for the vibration data came into being with the application of the turboset.With the rise of the process parameters and capacity, especially supercritical and ultra-supercritical generating units putting into operation, the more high reliability of the devices is pursued and the economic losses are also increased. Turboset has a lot of vibration data and thermal data, in the absence of corresponding information acquisition tool, when the fault occurs, we can not timely access to fault information. The above reasons propelled the development of the diagnosis technology matching the practice.The paper presents the study for the fault analysis and diagnosis system of the turboset vibration on the basis of the systematic reference and summary of the domestic and aboard turboset fault diagnosis development, related the characteristics of 900MW turboset and the characteristics of data collection system. We developed a combination of vibration parameters and thermal parameters of turboset vibration fault diagnosis system. The system has been applied to 900MW supercritical turboset. In the one year's steady operation, which was helpful to the safe operation of power plant.This paper pays emphasis on the following works:Firstly, The 900MW turboset of a power plant is the research object. This thesis introduces the general structure of the system. the combination of thermal and vibration parameters of the framework interface program was completed. Thermal parameters and vibration parameters of a joint diagnosis and recall function was completed.Secondly, the association rules based on the diagnosis modelis for supercritical units are put forward in this paper. It is the first time that a rule collection for specifical turbine shafting vibration diagnosis is built up with integration of thermal parameters,operation parameters and vibration parameters. The paper presents six main type faults through the summary for the former informant as well as the practical experience,particularly the fault type had found in 900MW units was classified and aggregated.The association rules mining algorithm is used for expressing the relationship between Signs of the vibration, heat parameter data and fault type using the association rules with confidence and support, constituted the rules of knowledge based on specific unit data's association rules mining.Then the collection is implemented with the program. With the reasoning and of the system, the fault symptoms and features are found out and the corresponding faults are identified.After the achievement of vibration fault diagnosis system based on association rules, it successfully worked out the causes of non-shaft vibration failure, and played a positive role.the practical operation showed that the system has good human-machine communication ability and diagnosis ability.It believed that Library association rules will be further improved and amended after a period of operation and experience. the diagnosis system coupling thermal and vibration parameters for power plant units will contribute to safe and economic operation.