Study On The Vibration Fault Diagnosis Of Steam Turbine Unit Based On Rough Set Theory

With the increasing capacity of steam turbine unit and the complexity of its structure, the large steam turbine unit demands higher usability, security, reliability and economic efficiency. Uncertainty and fuzzy problem of vibration fault for steam turbine is a bottleneck problem, and yet the reasons of causing vibration fault are diverse and complex. So, there are important theoretical significance and huge economic benefit for safe operation and studying on vibration fault diagnosis.Rough Set Theory is a mathematical tool of depicting uncertainty and imperfection. This paper puts forward a decision network model for vibration fault diagnosis, judging and diagnosing fault according to the tool. Firstly, discretize the original data of vibration fault, establish fault diagnosis decision table, proceed attribute reduction for decision table using discernibility matrix and improved discernibility matrix and obtain core attributes and reduction set of attribute. Secondly, we introduce confidence and support for rules set, rather than single concept of confidence, as well as thinking about the probability estimation of fault decision rule and the support of every decision rule. Not only are the amounts of rules set of fault decision decreasing substantially, but also is the matching efficiency of decision rule improving substantially. At last, reason the fault diagnosis network model and give the evaluation function of rules of network model. Check using examples to be patient and achieve the expected effect.Information entropy is metric of uncertain knowledge, which could depict the knowledge according to the certain numerical metric. Calculate the attribute information entropy and important degree, then reduce the fault diagnosis decision table. Results are the same using tow ways, at the same time, prove the accuracy and feasibility.The operation mechanism of model accords with logical reasoning of people, which was used neatly, and the diagnosis rules are simple, tidy and no-repeat. We can obtain satisfied results with confidence and support to diagnose the examples to be patient.