Nonlinear simulation of Kaplan turbine regulating system based on RBF networks

To simulate the hydroelectric power unit dynamic behavior as accurately as possible, the nonlinear characteristics of hydroturbine is considerable. This paper presents a nonlinear model of double regulated Kaplan turbine. Radial Basis Function (RBF) networks are applied to process the synthetic characteristic curve of Kaplan turbine. This improves the fitted data's accuracy. Based on this, the nonlinear Kaplan turbine regulating system model is developed in Simulink. The digital simulations of three cases for an actual hydroelectric power plant in China were performed, and the difference between nonlinear and linear Kaplan turbine regulating system model was analyzed. The results indicate that in small fluctuate transient process, the two models' simulation results are in good agreement, but in large fluctuate transient process, the nonlinear characteristics is not negligible and the two models' simulation results are deviate from each other seriously. The nonlinear model based on RBF networks is more accurate to reflect the regulating system's nonlinear dynamic process.