Vibration Fault Diagnosis Of Steamturbine Generating Unit Based On Rough Set And Multi-class Support Vector Machine

The paper has done a system research for the steam turbine vibration fault diagnosis’system architecture, the method use Rought set attribute reduction algorithm based on genetic algorithm and multiple support vector machine based on genetic algorithm research of steam turbine vibration fault diagnosis, research programs include the following contents:(1)Exploratory study of Steam turbine fault principle and signs, analysis and summary long-term accumulation of vibration fault features and fault symptom. Discusses in detail the common faults in rotating machinery, including mass unbalance of rotor misalignment, rubbing, oil film oscillation and fault mechanism and the vibration characteristics.(2)To improving the diagnosed speed and accuracy of the steam turbine diagnose system, this paper proposes a method that combines GA-SVM with Attribute Reduction on GA. Fristly we use Attribute Reduction on GA to the fault data preprocessing, cutting fuzzy fault symptom, extract and optimize rules, and then put the optimized decision table into GA-SVM, diagnosis. Examples prove that, compared with the traditional method, the proposed method can make the" error accumulation" phenomenon was significantly reduced, greatly improves the classification precision.