The Simulation And Fault Diagnosis Based On Neural Networks Of Hydro Turbine Governing System

Hydroelectric power has a very important position in China's energy production.Due to the large hydropower capacity in China and the ability of peak and phase adjustment of hydropower station,hydropower has received more and more attention in the development of power grids.Hydro turbine governing system is an important part in turbine controlling.The operating condition of the hydro turbine governing system will determine whether the entire power grid can operate safely and stably.In order to ensure the reliable operation of the hydro turbine governing system,it is necessary to monitor the status of the hydro turbine governing system as well as confirm and overhaul the faults that occur in the system timely.Now that there is a lack of fully functional condition monitoring and fault diagnosis system of hydro turbine governing system,this paper aims to develop a more practical condition monitoring and fault diagnosis system.This paper will focus on the modeling and fault diagnosis based on neural networks of hydro turbine governing system.This paper studies the mathematical model of hydro turbine governing system.The MATLAB software is used to model the hydro turbine governing system.Through the noload disturbance experiment and load disturbance experiment in the simulation,the correctness of the simulation model is verified.In order to transplant the model into the platform of this system,this paper puts forward a method named discrete analog by links which is suitable for the simulation of hydro turbine governing system.This method can efficient achieve the simulation process in the C++ Builder platform by dividing transfer functions into links and finding the system's contact matrix.At last,the comparison of simulation results verifies the correctness of the method.This paper studies and implements fault diagnosis based on BP neural network.After that,through the research on the problem of coordination of neural network diagnosis method and expert system diagnosis method,this paper proposes a new cooperation scheme of neural network diagnosis method and expert system diagnosis method.This scheme uses fault symptom membership degree and symptom judgment matrix to judge the credibility of new knowledge that may appear in neural network diagnosis,thus the problem of how to use neural networks to acquire expert knowledge is solved.Based on the above researches,this paper uses C++ Builder software as a development platform and SQL Server as a database software to develop the system software that has the functions in the design plan.