中文摘要：

区间振荡是大规模交直流互联电网安全稳定运行面临的主要问题之一。文章将一种新型自适应动态规划算法——Gr HDP（goal representation heuristic dynamic programming）算法应用于设计柔性直流输电的附加阻尼控制器（supplementary damping controller,SDC）以阻尼区域间功率振荡、提高交直流互联系统的整体动态稳定性。Gr HDP是在HDP（heuristic dynamic programming）2层神经网络的基础上发展而来的,它增加了一个目标网络,能形成自适应的内部强化信号,来调节输入信号与输出控制量之间的映射关系,从而大大改善系统的动态性能。Gr HDP-SDC相对于常规超前滞后SDC的优点是,前者的设计无需精确的被控对象模型,且对系统运行状态的改变具有很强的学习能力和适应性。以含两端柔性直流输电的四机两区域系统为例进行了仿真研究,仿真结果表明,Gr HDP-SDC与常规超前滞后SDC、HDP-SDC相比,能更好地改善系统的动态性能,且能更好地适应运行点的变化和不同的扰动情况。

英文摘要：

Inter-area oscillation is a main challenge for secure and stable operation of large-scale interconnected AC/DC power system. In this paper, an innovative adaptive dynamic programming approach, namely goal representation heuristic dynamic programming（Gr HDP）, was proposed to design supplementary damping controller（SDC） of voltage source converter HVDC（VSC-HVDC） transmission system to suppress inter-area oscillation in large-scale AC/DC power system. Based on the two neural networks structure of HDP, Gr HDP developed a new goal representation network, able to automatically generate internal adaptive reward signal to facilitate better mapping between system state and control action. Therefore, Gr HDP-SDC could significantly improve dynamic performance of power system. Not necessarily knowing exact mathematic model of the power system, Gr HDP-SDC possessed quick learning and universal control characteristics and strong adaptability, superior to conventional lead-lag SDC. Case study was performed based on a two-area four-machine power system with VSC-HVDC transmission line. Conventional lead-lag SDC and HDP-SDC were also studied for comparison. Simulation results show that the proposed Gr HDP-SDC has better performance in damping inter-area oscillation than that of conventional lead-lag SDC and HDP-SDC in a wide range of system operating conditions.