Research On Key Technologies Of Condition Maintenance System For Hydropower Generator Units

According to condition based maintenance can mostly prolong the maintenance intervals and reduce operation cost, it can ensures safety operating of hydropower plants and the successive development of national economy. Nowadays, maintenance model is transferring from plan maintenance to condition based maintenance step by step. Since hydropower generator unit is a very complicated system that involving hydraulic, electrical and mechanical factors, such as the fault mechanism, diagnosis methods, data channel, evaluation principle of life and maintenance decision supporting system are not fully researched. Condition based maintenance system focuses on such fields as reducing units'maintenance cost, improving devices'operating efficiency and maintenance predictive. Since it is a difficult engineering includes electricity fields and management fields, there is no actual condition based maintenance system in real life. In line with performance test and monitoring diagnosis project that research team responded for, such key technologies for condition based maintenance as: fault mechanism, diagnosis methods, data channels, signal extracting, symptom identification and trend-based forecasting. All of these research can do some good for the eventually implementation of condition based maintenance. Fault reasons analysis is essential for fault diagnosis. It does some research on the extraordinary runner hang-up in Tian Huangping Hydropower Station, the abnormal vibration of Three Gorges Hydropower Station left bank #6 generator unit under 4% guide vane opening and the mechanism of turbine runner crack. By means of theoretical analysis, numerical simulation and prototype test. Research results indicate: the main cause of hang-up is the variety of leakage discharge of sealing ring, which results in the change of axial force of shaft, and the jam or damage of seal parts may lead to the leakage discharge variation; "negative discharge"can not be the cause of abnormal vibration under small opening and possible reason of "fluid structure resonance"is also analyzed; when static and dynamic stress acted on runner blade reach fatigue limitation, accumulated fatigue damage may leads to the micro crack or extend of crack, is responsible for runner blade fracture, and excessive dynamic stress is the main cause. And traditional diagnosis methods are not adequately. CBR (Case Based Reasoning) is introduced into hydropower generator sets fault diagnosis, and it lowers the difficulty of knowledge acquisition; also shafting model of hydropower generator units is modified to relief the bottleneck of knowledge repository by implementing shafting simulation results as the fault diagnosis knowledge database. Since oil film force acquisition by interpolating the oil film force database, time complexity is reduced. A diagnosis strategy based on multi method diagnosis league is also introduced to improve the diagnosis accuracy. Equipments'operating conditions can be obtained through axis orbit and conventional identification algorithm of axis orbit is low accuracy or too complex. A new method of axis orbit automatic identification base on moment invariant is proposed. It acquires the axis orbit shape parameters to help operation personnel to deeper diagnosis by ways of optimization method. Since signal extract algorithm is critical to determine whether symptom reflects current operating conditions and signal trend-based forecasting algorithm effects the decision of future operating conditions, local weak signal is extracted based on harmonic wavelet and prediction in line with maximal Lyapunov exponent is introduced into hydro-generator unit fields Data channels are also analyzed: a hybrid network transmission model of B/S (Browser/Server) and C/S (Client/Server) is used to consist of distributed monitoring and diagnosis system; memory database is adopted to decrease the times of operation of hard disk and improve the real–time performance; a new intelligent data store algorithm is brought forward to reduce the data size that stored in database and also assures sufficient data to analyze; Matlab Web Server is the first time introduced into hydro-generator sets faults diagnosis to implement distributed remote monitoring and diagnosis system.