一种电力系统网络化预测阻尼控制方法及其关键技术研究

The WAMS-based wide-area damping control of power system is used as the research object. By introducing the networked predictive control theory, a networked predictive damping control strategy for power system is proposed in order to deal with the problems brought from characteristics of the power system, such as time-varying, random delays, multi-variable, strong constraints, et al. According to the key issues of applications of this proposed control strategy in power system, the following four issues is focused on: wide-area signals and the control points selection, the low-order state space predictive model online identification, control sequence generation and the network delay compensation strategy. Firstly, by combing of the geometric method and the relative gain array method, a wide-area signals and control point selection method, which can adapt to the variation of system operating condition, is proposed. Secondly, based on online subspace identification theory, an online identification algorithm is proposed for identifying the low-order state space model of the power system in order to adapt the change of operation conditions; Thirdly, the objective function and control parameters is selected based on the actual damping control requirement, fast and efficient predictive control generation algorithm is used to generate the predictive control sequence; Finally, a network delay compensation strategy is studied to compensate the constant and random communication delay effectively. The networked predictive control based wide-area damping control design theory will be formed. The research results are expected to explore a new way to achieve wide-area damping control in practice power systems.

以基于WAMS的电力系统广域阻尼控制为对象，引入网络化预测控制理论，提出了一种电力系统网络化预测阻尼控制的新方法，以解决电力系统中运行状态复杂多变、随机延迟、多变量、强约束等特性给控制系统带来的问题。针对该方法在电力系统中应用的关键问题，着重从广域信号和控制点选择、预测模型在线辨识，预测控制序列产生策略和网络延迟补偿策略这四个方面展开分析研究：结合几何方法和相对增益矩阵方法，提出了能适应系统运行状态变化的广域信号和控制点选择方法；研究在线递推子空间辨识理论，提出能够在线辨识电力系统低阶状态空间模型的辨识算法以适应系统的运行状态变化；根据实际阻尼控制要求选定有效的目标函数和控制参数，选择快速高效的预测控制算法来产生预测控制量序列；研究并提出网络延迟补偿策略以有效补偿固定和随机时延。最终形成一套基于网络化预测控制的广域阻尼控制设计理论，研究结果可望为实现电力系统广域阻尼控制探索出新的途径。

采用广域测量信号设计广域阻尼控制器（WADC）可以有效增强大规模互联电力系统的关键区间振荡模式的阻尼，保证其安全稳定运行。广域测量信号通常采用通信网络进行传输，因此在传输过程中不可避免会引入随机通信延迟，从而会恶化控制器的阻尼效果甚至导致不稳定。针对上述问题，项目组首先分别采用基于自由权矩阵的单时滞稳定判据和多时滞稳定判据计算含单个WADC和多个WADC的电力系统的时滞稳定裕度，并将计算所得的时滞稳定裕度作为控制器附加设计指标来设计考虑时滞影响的WADC。典型算例仿真研究表明，所设计的WADC不仅能很好地阻尼系统区域间低频振荡，还具有较强的时滞鲁棒性。由于WAMS环境下的WADC设计属于网络化控制系统(NCS)理论研究范畴，因此可以采用NCS控制理论来解决广域信号传输延迟影响的问题。项目组提出一种基于网络化预测控制(NPC)的WADC的设计方案，所采用的NPC是在广义预测控制(GPC)的基础上，增加NDC以补偿固定和随机通信延迟。用带变遗忘因子的递归最小二乘法在线辨识获得低阶预测控制模型，以克服系统运行方式复杂多变和系统参数不精确引起的控制性能下降的不利影响。项目组将NPC算法分别用于设计发电机励磁控制的WADC和静止无功补偿器(SVC)的广域附加阻尼控制器，典型算例仿真结果表明，所提出的基于NPC的WADC不仅能够有效阻尼区域间的低频振荡，还能主动补偿固定和随机通信延迟的影响和适应系统运行状态的改变。此外，还将NPC理论扩展应用到多区域电力系统的负荷频率控制中，提出了一种基于递归型NPC和模型辨识的分散负荷频率控制器，并仿真验证了所提控制器的有效性。