System Identification Based Optimization Of Control Parameters For Hydro Turbine Generating Units

Hydro turbine generating units play important roles in the power system. With the improvement of power quality requirements, it's needed to improve control performance of the governor. In addition, the traditional parameters tuning method of the governor may waste a lot of water resources. To resolve the problem, the model of the hydro turbine governing system is identified first and then the control parameters of the governing system is optimized based on the model identified above.Least square (LS) identification method is mature and has very sound theoretical consistency and convergence. It is used widespread and can be easily developed with a computer. It becomes one of main methods of system identification. Particularly, recursive least square (RLS) is suitable for on-line real-time identification. So RLS is chosen to identify the system.Ant colony algorithm (ACA) is proposed to search the optimal control parameters of the governing system based on the model identified above. ACA is positive feedback and robust. Improved ACA, combined with model reference adoptive control, is used to optimize the control performance of the governing system.Simulation model is built based on Matlab with Simulink. ACA and genetic algorithm (GA) programs are developed on the platform of Matlab. A series of identification and optimization simulation experiments are performed based on above work. Pseudo-Random Binary Sequence (PRBS)signal is chosen as excitation of the identification experiments.The simulation experiments demonstrate that the difference between the identified model and the real one is rather small. The method is simple, effective and easy to implement. The optimization of the control parameters can be further performed based on the identified model to improve the dynamic performance. The comparison among the proposed ACA, GA and fminsearch which is the function of MATLAB toolbox is made. The result demonstrates that the time consumed with ACA optimization is least. Therefore it is valuable and practical.