Multi-Agent System Based Special Protection and Emergency Control Scheme against Cascading Events in Power System

This thesis concerns the development of wide area special protection and emergency control scheme that can provide effective countermeasures against long term voltage instability induced cascading events and blackouts in power system. Most past cascaded blackouts are caused by unexpected backup relay operations due to low voltage or overload state in the post stage of N-1 (or N-k) contingency. If such state could be sensed and adjusted appropriately before those relay actions, the system stability might be sustained. So it is of great significance to develop a suitable protection scheme to identify and adjust those emergency states in advance before a cascaded blackout.

In order to identify the diverse emergency states in the post fault stage, the distributed distance relays with upgraded settings which take the capability limits into account will be utilized in this thesis. Moreover, according to the variations of power system operation, an overload prediction algorithm based on impedance sensitivity is proposed here to distinguish the emergency states from a remote fault. These methods combined together, which can give a fast identification of the emergency states and a clear mapping of critical relays in the post fault stage.

For the purpose of adjusting the identified emergency states timely and preventing the unexpected relay operations, a fast optimal control algorithm is proposed here to find the most effective control locations and reasonable control amount, which is based on sensitivity analysis and the extent of specific emergency states. At the same time, load restoration dynamics is taken into account to compensate sensitivity calculation and correct the control algorithm. Such control methods provides a direct and simple way from problems to solutions, which is suitable for real time implementation.

On the other hand, with the aim of executing the control methods timely and implementing the whole protection strategy suitably, multi-agent system (MAS) with hierarchical structure is designed here. The distributed relays and controllers, which work as device agents in the lowest level of MAS. And the agents in higher levels will be used to build an effective cooperation environment. Therefore, those distributed agents can be systematically organized by MAS to effectively achieve the global goal on preventing the post fault emergency states and subsequent cascading events.

Moreover, in order to handle some unexpected failures of control methods’ execution or some unexpected disturbances when the power system is already operated in an emergency situation, the process control strategy is designed in this thesis to manage the execution progress of whole protection strategy in real time. If there is an unexpected situation emerging, the agents in higher level will be activated to solve this problem. Such methods improve the reliability of the proposed protection strategy coping with more complex emergency situations than planned.

Finally, in order to realize and testify the proposed protection strategy in this thesis, a real time simulation platform based on Real Time Digital Simulator (RTDS) and LabVIEW is built. In this platform, the cases of cascaded blackouts are simulated on the test system simplified from the East Denmark power system. For the MAS based control system, the distributed power system agents are set up in RTDS, while the agents in higher level are designed by LabVIEW toolkits. The case studies and simulation results demonstrate the effectiveness of real time application of the proposed MAS based special protection and emergency control scheme against the cascaded blackouts.