Modeling And Simulation Of Neural Network For Regulation Synthetic Digital Characteristics Of Hydro-Turbine

Extracting reasonable flow and torque characteristics data from the hydro-turbine synthetic characteristic curves has decisive significance for the turbine selection and distribution system regulate calculation, which is the effective protection for the station's safe and economic operation. At present, there are many hydro-turbine synthetic characteristic curves treatment, but most of these only stay in high efficient region. To get the characteristic data of the low efficient region remain depending on the designer's experience to extend the curves, which need repeat manual test. As the turbine has multiple nonlinear regulation characteristic in the low efficient region, the data which was obtained by the artificial curve extension can not real reflect the hydraulic flow and torque regulation characteristic. Improved BP neural network has a strong nonlinear curves treatment function to obtain high precision forecasting data. Combine the known discrete data and boundary constraint to obtain the simulation surface of flow extension and torque extension in the low efficient region. Summed up the whole process of the research, mainly get the following several aspects of content:(1)Comparing several commonly using hydro-turbine synthetic characteristic curves treatments, combining the hydro-turbine's actual curve parametric characteristics, chose the cubic spine interpolation as the curve treatment in high efficient region.(2)Using the Computer Aided Design (CAD) to make Digital Processing for the grating image of hydro-turbine synthetic characteristic curves, obtained the flow and torque regulation characteristic data in high efficient region.(3) Combined the neural network and Matlab establish at three-tier structure improved BP neural network, and used the known discrete flow and torque data to train the neural network, through the Matlab model simulation and error analysis, verified that the model was Very accurate.(4)Taking advantage of the curves data points'extension technology, combining the boundary constraint and the improved BP neural network's predicting function, got the simulation surface of flow extension and torque extension in the low efficient region.(5)Comparing the flow and torque predicting characteristics data with the engineering test results, verified that the hydro-turbine synthetic characteristic curves improved BP neural network model can reflect the real regulation characteristics.Using improved BP neural network to treatment the low efficient region hydro-turbine synthetic characteristic curves is a new model of nonlinear modeling and simulation model for researching the hydro-turbine control system, which can real reflect the regulation characteristics of the hydro- turbine.