Research On Model Predictive Emergency Voltage Control Of Power Systems Based On Trajectory Sensitivity Technology

After a power system is subjected to a disturbance, how to implement the emergencyvoltage control to effectively prevent from voltage collapse or collapse of the power systemwas an important defense line of security and stable operation of power system. The paperaiming at long-term voltage stability scene, based on quasi-steady-state model, centralizedmodel predictive control and distributed model predictive control are introduced the field ofemergency voltage control by means of employing trajectory sensitivity technique of hybridsystems. Ultimately it can achive the online application for the model predictive emergencyvoltage control system.The centralized model predictive emergency voltage control model of the whole systemwas established. Minimizing both the voltage magnitude deviation of the buses and controlcost of various control schemes, from a single load-shedding to comprehensively consideringcontrol means of a variety of stable voltage, such as generator AVR setpoints, switchablecapacitor banks, on load tap changers and load shedding, regarding the both weighted value asan objective function, the centralized model predictive emergency voltage control model of thewhole system which regarded differential-algebraic equations with both continuous anddiscrete variable as restraints was established in order to enhance the voltage stability of thepower system. By employing trajectory sensitivity techniques and quasi-steady-statesimulation, it can be obtained the trajectory sensitivity of state variables and algebraicvariables relative to the control variables, so the centralized models predictivevoltage-controlled model was simplified. Thus it can improve the efficiency of calculation ofrolling optimization function.The paper proposes the communication based distributed model predictive control(CDMPC) and cooperative distributed model predictive control (CODMPC) in emergencyvoltage control. In CDMPC grid system is divided into several sub-regions. Through thedecomposition of inter-regional coordination of technology to deal with the border issue, thesystem model and controller are decomposed. Each regional establishing MPC scroll objectivefunction, the system-wide Nash optimal solution can be obtained under networkcommunication between these regions. On the basis of CDMPC, CODMPC's objectivefunction of sub-regional is expressed with convex combination of all sub-regional' objectivefunctions, which uses trajectory sensitivity techniques and network communication to predicttrajectory of voltage, then its goal cost function is changed into quadratic programmingfunction with all control variables. So it avoids regional optimization conflicts each other. Two DMPC methods are used to simplify a large-scale centralized optimization problem bybreaking them down into many smaller ones, thereby the size and complexity of the problemis reduced greatly.MPC possesses some features such as no requirement of precise mathematical model ofthe system and convenience in processing all kinds of operational constraints andmulti-variable issues, which are excellently suit to power system voltage stability. Non-linearobjective optimization function is simplified as a linear combination of the cost functions bytrajectory sensitivity technology, this improves the optimization efficiency. Compared with thetraditional optimal coordination of voltage control, simulation examples show that theproposed methods can effectively enhance the system voltage stability, and improve thepracticality of the model predictive control algorithm.