Modeling And Control Of Power System Based On Goal Representative Adaptive Dynamic Programming

Power system modeling is the basis of its analysis, calculating, control. It is very difficult to accurately model because of various operating conditions and uncertain parameters. On the other hand, modern power system is a non-linear, high-dimensional, large-scale dynamic system. Appropriate and effective control is the key to ensure security and stability of power system. Since the existing controllers are designed based on a particular point in power system, it is difficult to ensure good control effect, even can't keep the system stable in a variety of complex operating conditions. Multi-level heuristic dynamic programming has strong adaptability. It gradually learns to master the dynamic characteristics of system by offline training and online correction. When the model is unknown or the parameters are uncertain, the algorithm can also provide a model or control. In this paper, a multi-level heuristic dynamic programming is used in power system modeling and control. The main contents are as follows:In this paper, the multi-level dynamic programming algorithm is studied in depth on the basis of heuristic dynamic programming. The simulation of inverted pendulum system show that the controller designed based on GrHDP is more rapidly in tracking for optimal target than HDP-based controller. GrHDP-based controller is also faster to stable and has more precise control effect.For the current high voltage problem of point of common coupling, the paper proposes a voltage reactive power control strategy of static var compensator based on GrHDP algorithm. The simulation in 4-machine 2-area system shows that compared with the SVC voltage reactive power controller with PID designed by particle swarm optimization, the SVC voltage reactive controller based on GrHDP can more effectively control the overvoltage of the point of common coupling under different operating conditions and fault conditions.For the response of dynamic equivalence unprecise after changes in power system, the paper proposes a dynamic equivalent method based on GrHDP algorithm. The simulation in the 10-machine 39-bus system shows that the dynamic equivalent method proposed based on GrHDP algorithm can more precisely trace the dynamic properties of the original system under different operating conditions and fault conditions.