| With the development of hydropower state monitoring technology, the operational status of hydropower units to conduct a comprehensive monitoring, real-time health status and abnormal fault hydropower units for fast and accurate predictive analysis, is essential for the protection of crew and grid stability and security. Therefore, the hydroelectric generating set maintenance mode based on the state of maintenance is the current trend in the field of research where the predicted trend for the state unit equipment is an important part of the implementation of condition-based maintenance,state-based maintenance mode state has a good role in promoting early warning unit.In this paper, the prediction method of neural network and support vector machine(SVM) are specific researched, and both neural networks and support vector machines methods are applied to the hydroelectric generating Predicting Condition. Since the process of hydropower units run susceptible to various factors such as the the mechanical aspects,the electrical aspects and the hydraulic aspects of the terms and many other factors, so the fault type has a variety of features, and the vibration of hydropower units fault is the most common and important one. Therefore, we selected two vibration RMS vibration for the type of fault, both RMS vibration of hydropower units can reflect the operational status and future operating trends, then establish BP neural network of prediction model and support vector regression(SVR) of prediction based on the practical application of engineering according to historical values acquired, and predict the effect of the error analysis.Through practical test, the results show that support vector machine forecasting method for vibration parameters of the prediction is better than that of BP neural network prediction method, it will also provide a reference basis for subsequent fault diagnosis, and the security is of great importance to the safe and stable operation of unit. |
PDF Full Text Request