Weighted multiple model adaptive LQG and PI control for a hydro turbine plant

Hydro turbines are known to be highly nonlinear plants with uncertainties especially in the presence of large or abrupt disturbances. The presence of some nonlinear components within the system also increases its complexity. In this study, weighted multiple model adaptive control (MMAC), is proposed to be used as a controller for such systems, with the aim of providing desirable characteristics such as robustness and stability. To assure stability, the system was designed in accordance with the virtual equivalence system and adaptive mixing control criteria. Two different control structures, namely the LQG/LTR and PI controllers were optimized to meet these requirements and tested through simulations. For each control structure, four different linear compensators were tuned according to specific percentage load disturbances, and each compensator corresponds to an a priory identified candidate linear model of the plant. The LQG/LTR and PI controllers were designed for robustness and blended into the MMAC scheme. The plant operation, subjected to different disturbance conditions, was simulated. From the obtained results, the LQG/LTR scheme exhibited good performance for smooth disturbance changes but fluctuated in situations with sudden changes. The PI controller showed good performance for both, smooth varying and sudden applied disturbances.