GPU并行的电力系统连锁故障快速仿真算法

GPU Parallel Fast Simulation Algorithm for Cascading Faults in Power Systems

针对各类极端灾害频发、电力系统发生大停电的风险问题,该文研究了电力系统连锁故障的演变路径,对降低停电损失、提高灾后恢复能力具有重要意义。基于仿真模拟,以极端灾害导致的故障元件作为初始故障集,利用GPU并行静态安全分析,获取初始故障发生后系统的下层风险元件;利用GPU并行潮流计算模块对严重故障集进行交流潮流分析,计算线路故障的负载率、耦合指标、潮流波动指标和电压越限情况等关联性指标,判断出最严重的故障情况作为下级故障元件,当电压指标越限时采用最优潮流并行模块计算最小切负荷;根据故障链结束判据终止计算,得到初始故障集下的连锁故障演变路径。通过39节点系统算例进行分析和验证,说明了算法的有效性和加速效果。

In view of the frequent occurrence of various extreme disasters and the greater risk of power system blackouts,it is of great significance to study the evolution path of power system cascading faults in order to reduce the loss of blackouts and improve the ability of post disaster recovery.Based on the simulation,the fault elements caused by extreme disasters are chosen as the initial fault set,and the lower risk elements of the system after the initial fault are obtained by GPU parallel static safety analysis,and then the AC power flow analysis of the serious fault set is carried out by GPU parallel power flow calculation module to calculate the load rate,coupling index,power flow fluctuation index,voltage out of limit and other correlation indexes of the line fault,and judge the most serious fault condition as the lower fault element. If the voltage index exceeds the limit,the optimal power flow parallel module is used to calculate the minimum load shedding in addition to disconnecting the line. Finally,according to the termination criterion of the fault chain,the evolution path of cascading faults under the initial fault set is obtained. An example of 39-bus system is analyzed to verify the effectiveness and acceleration effect of the algorithm.