Vibration Adaptive Anomaly Detection of Hydropower Unit in Variable Condition Based on Moving Least Square Response Surface

摘要

It is difficult to effectively analyze and identify the conditions of hydropower unit, due to its complex operation conditions, frequent start-stop conditions, continual working status switch, less fault samples, single static alarm threshold. Lots of test research shows that active power and working head are key factors which affect the operation conditions of hydropower unit. The health standard condition of unit is determined. An adaptive real-time anomaly detection model of hydropower unit vibration parameters is proposed based on moving least square response surface. In the proposed model, active power and working head are comprehensively considered. This model can adapt variable conditions of hydropower unit. The model is used to real time detect the anomaly of hydropower unit vibration parameters. The results show that this model can effectively evaluate the performance of unit vibration, can more accurately detect the abnormal of unit vibration.