基于轨迹灵敏度的协同分布式模型预测电压控制

Cooperatively-Distributed Model Predictive Control of Voltage Based on Trajectory Sensitivity

为克服集中式模型预测电压稳定控制的局限性,基于系统分区和预测电压控制模型分解提出了一种适于大型电力系统的协同分布式模型预测电压控制方法.该方法以所有子区域目标函数的凸组合来构建全局目标代价函数,利用轨迹灵敏度技术预测节点电压变化轨迹,将目标代价函数转化成与所有子区域控制变量有关的二次函数,从而避免了各子区域之间目标优化可能存在的冲突.在两个试验系统上的仿真结果表明,文中方法能获得与集中式模型预测电压控制相近的效果,且计算效率显著提高.

In order to overcome the limitations in voltage stability of the centralized model predictive control, a cooperatively-distributed model predictive control method based on the system partition and the predictive control model decomposition of voltage is constructed for large-scale power systems. In this method, a global objective cost function is proposed with the convex combination of the objective functions of all subareas, the nodal voltage trajectory is predicted according to the trajectory sensitivity, and the objective cost function is transformed into a quadratic programming function relevant to all control variables, thus avoiding the optimization conflicts among the subareas. Simulations on two test systems indicate that the proposed method greatly improves the calculation efficiency and achieves a voltage control performance similar to that of the centralized model predictive control.