This paper applies the ultra-local models (also labeled as intelligent PID controllers) to Automatic Generation Control (AGC) in a Multi-Area power system (New England 39 Bus) in the presence of communication delay. Given the fast development of smart grid in the past years, power system and communication network are coupled much more tightly than before. Therefore, the independent study of the two domains or simplistic modeling the communication delay in the power system simulation environment is no longer effective and convincing. In our work, the power system model containing AGC is built in Simulink/Matlab, and the communication network is co-simulated in Network Simulator 2 (NS2) through PiccSIM, a simulation platform for (wireless/wired) networked control systems.

A careful comparison is made between the performance of intelligent-P (iP) controller and the conventional PI controller under the co-simulation environment of AGC with time-varying wired communication delay. The results show the two controllers are largely equivalent for relatively small frequency deviations. However, intelligent-P controller performs better than conventional PI controller under more severe system variations, such as large communication delay. The avoidance of integration in the intelligent-P controller makes anti-windup algorithm unnecessary, which enhances the applicability and the simplicity of controller tuning.

In addition, the co-simulation results also demonstrate that the system is more sensitive to the communication delay of control signals from the area control center to the distributed generators than to the delay of tie-line active power measurements. This conclusion is very illuminating that the distributed control mechanism becomes a promising alternative for the control mechanism applied widely now. Instead of a one control center per regulating area, every generator participating in AGC could have its own local controller. The distributed generator controllers can communicate with each other and share the needed information. This fully-distributed automatic generation control (AGC) is more tolerant of communication delay, and thus more robust and stable under various conditions.