MPC-Based Control of a Large-Scale Power System Subject to Consecutive Pulse Load Variations

One of the classic approaches to controlling power networks, as large-scale systems, has been the use of a centralized control architecture. This approach is currently used less frequently due to its computational complexity. Another possible approach is the use of a decentralized control architecture. However, this approach can lead to unacceptable global performance of the system due to the lack of knowledge about the available interactions among subsystems. A third approach is the application of a cooperatively distributed architecture. On the other hand, one technique that has proved to be quite efficient for the control of power system frequency is model predictive control (MPC). In this paper, the performance of cooperatively distributed MPC is compared with that of the centralized MPC and the classical automatic generation control methods. The main contribution of this paper is that the load variations are applied to the system in the form of consecutive pulses. Additionally, the disturbance levels considered here have higher values. Moreover, the range of control input variations is reduced; therefore, the constraints are chosen more strictly. Finally, the total error of the system is determined, and the discussed methods are evaluated by the newly defined indices. According to simulation results, a feasible cooperation-based MPC method leads to relatively desired performance and computational speed and so can be an appropriate practical option for controlling power systems.