基于多层次自适应动态规划的电力系统建模与控制

发布时间：2018-11-04 16:44

【摘要】：电力系统建模是对其分析、计算、控制的基础。由于系统运行状态多变及参数的不确定性,进行电力系统精确建模非常困难。另一方面,现代电力系统是非线性、高维度、大规模的动态系统,对其采用恰当有效的控制是保障电力系统安全稳定运行的关键。现有的电力系统控制器往往基于某一个运行点设计,难以保障在系统各种复杂运行工况下保持好的控制效果,甚至不能保持系统稳定。多层次启发式动态规划具有很强的自适应性,通过离线训练和在线修正,逐渐学习并掌握系统的动态特性,在模型未知和参数不确定的情况下,能够进行电力系统建模和控制。本文将多层次启发式动态规划用于电力系统建模和控制,主要研究内容如下:在启发式动态规划(Heuristic Dynamic Programming,HDP)的基础上,深入地研究多层次启发式动态规划(Goal representative Heuristic Dynamic Programming,GrHDP)算法。以倒立摆系统为例进行仿真研究,结果表明:基于GrHDP设计的控制器比基于HDP设计的控制器对最优目标的跟踪更迅速,稳定速度更快,控制效果也更加精确。针对目前风电场并网点处静止无功补偿器(Static Var Compensator,SVC)装置控制系统带来的高电压穿越问题,提出了基于GrHDP的SVC电压无功控制策略。以4机2区域系统为例进行仿真研究,结果表明:在不同的运行工况和故障情况下,与采用粒子群优化算法设计的PID型SVC电压无功控制器相比,本文所提的基于GrHDP的电压无功控制器都能更有效抑制故障后风电场并网点的过电压。针对电力系统运行工况变化后动态等值系统响应不够精确的问题,提出了基于GrHDP的电力系统动态等值算法。以10机39节点系统为例进行仿真研究,结果表明:在不同运行工况和故障方式下,相比于同调等值法,本文提出的基于GrHDP的电力系统动态等值算法都能更精确地跟踪原系统的动态特性。
[Abstract]:Power system modeling is the basis of its analysis, calculation and control. The accurate modeling of power system is very difficult because of the variable operation state and uncertainty of parameters. On the other hand, modern power system is a nonlinear, high-dimensional, large-scale dynamic system. Proper and effective control is the key to ensure the safe and stable operation of power system. The existing power system controllers are often designed on the basis of a certain operating point, which is difficult to guarantee the good control effect under various complex operating conditions of the system, or even to maintain the stability of the system. Multi-level heuristic dynamic programming has strong adaptability. Through off-line training and on-line modification, the dynamic characteristics of the system can be gradually learned and mastered, and the power system modeling and control can be carried out under the condition of unknown model and uncertain parameters. In this paper, multi-level heuristic dynamic programming is applied to power system modeling and control. The main research contents are as follows: on the basis of heuristic dynamic programming (Heuristic Dynamic Programming,HDP), the multi-level heuristic dynamic programming (Goal representative Heuristic Dynamic Programming,) is deeply studied. GrHDP) algorithm. Taking inverted pendulum system as an example, the simulation results show that the controller based on GrHDP can track the optimal target more quickly, the stability speed is faster and the control effect is more accurate than the controller based on HDP. Aiming at the problem of high voltage traversing caused by the control system of static Var compensator (Static Var Compensator,SVC) at the point of wind farm, a SVC voltage and reactive power control strategy based on GrHDP is proposed. The simulation of 4-machine 2 area system is carried out. The results show that under different operating conditions and fault conditions, compared with PID type SVC voltage and reactive power controller designed by particle swarm optimization algorithm, The voltage and reactive power controller proposed in this paper based on GrHDP can restrain the overvoltage of wind farm after fault more effectively. A dynamic equivalent algorithm based on GrHDP is proposed to solve the problem that the dynamic equivalent system response is not accurate enough after the change of power system operating conditions. The simulation of a 10-machine 39-bus system is carried out. The results show that under different operating conditions and fault modes, compared with the homology equivalent method, The power system dynamic equivalence algorithm based on GrHDP proposed in this paper can track the dynamic characteristics of the original system more accurately.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TM732

【参考文献】