On Data Fusion Fault Diagnosis And Simulation Of Hydroelectric Sets Vibration

Vibration of hydroelectric sets is main parameter that reflects the running status of the sets. It is vital to monitor vibration and diagnose the vibration fault of hydroelectric sets, and monitoring and fault diagnosing is important measure to ensure running safely and stably of hydroelectric sets.According to the present situation and developing trends on condition monitoring and fault diagnosing of hydroelectric sets at home and abroad, and taking the technique of data fusion diagnosis hydroelectric sets as theme, this dissertation researches into data fusion fault diagnosis method based on Dempster-Shafer theory, and applies this technique to hydroelectric sets vibration fault diagnoses for the first time, and presents that using sub-band energy of vibration frequency to reflect the vibration intensity of hydroelectric sets. What is mainly discussed in this dissertation is as follows:The fault mechanism and vibration frequency feature of hydroelectric vibration are analyzed, induced and reorganized. The select and placement of detecting positions of parameters such as vibration, swing and pressure pulsation are researched, and the corresponding sensors are selected tomonitor.Based on the discussion of the base principle and several fusion diagnosis methods of multi-sensor data fusion, data fusion diagnosing method based on D-S theory is mainly researched; According to the requirement of fusion principle, constitute methods of mass function are discussed, and the identify frame for fusion diagnose of hydroelectric sets vibration fault are constructed.In simulation, several position vibration signals of hydroelectric sets are simulated, and their features are extracted with Wavelets and Fourier theory. Then simulation diagnosis is made using data fusion method.The result shows that fault diagnosis based on multi-sensor data fusioncan make the most of the information of multi-sensor in hydroelectric sets positions, decrease diagnosis uncertainty, and identify faults which cannot be recognized with single sensor data, and that fault without obvious early features can be found as soon as possible and identified accurately by means of multi-sensor data fusion.With the increasing of fusion parameters, the reliability of diagnosis can be improved. It is no doubt that it is vital for find underlying fault at an early date, and it is significant for ensuring running safely and stably of hydroelectric sets and realizing condition overhaul of sets.