Simulated response of NN based identification and predictive control of hydro plant

This paper studies the neural network nonlinear autoregressive with exogenous signal (NNARX) model identification of elastic and inelastic hydro power plant. A nonlinear relationship between the turbine deviated power and random gate position on random load variation and water disturbance is assessed. The identified elastic NNARX hydro plant model is simulated with predictive controller to track a given deviated power as a reference signal. The controller parameters are optimally determined by solving quadratic performance index using well known Levenberg–Marquardt and quasi-Newton algorithm. And it is demonstrated that the deviated power tracks its deviated power target signal accurately over wide rapid-variations in load and water disturbances.