基于CNN-LSTM网络的电网电压稳定紧急控制策略

Emergency Control Strategy of Power Grid Voltage Stability Based on Convolutional Neural Network and Long Short-term Memory Network

运行方式的复杂多变性、扰动故障的不确定性、电力电子设备的弱抗扰性为交直流混联电网的安全稳定运行带来了巨大挑战。为保证大扰动故障后电网电压的稳定,提出一种基于卷积神经网络(CNN)和长短期记忆(LSTM)网络的响应驱动紧急控制策略。首先,分析关键母线节点电压时序值和电压稳定水平的映射关系,离线建立基于CNN-LSTM网络的大干扰电压稳定评估模型。然后,采用评估模型预测备选切机、切负荷点控制措施动作后电网电压稳定水平的提升量,确定响应驱动紧急控制措施灵敏度。最后,考虑紧急控制措施灵敏度,建立计及电网实际运行约束的紧急控制优化问题,求解得到最优紧急控制策略。面向存在电压失稳问题的交直流混联电网实际场景,仿真结果验证了所提响应驱动控制灵敏度预测方法的准确性,电压稳定紧急控制措施优化协调策略可以保证大扰动故障后电网的安全稳定运行。

The complexity and variability of operation mode,uncertainty of disturbance fault and weak immunity of power electronics equipment bring great challenges to the safe and stable operation of hybrid AC/DC power grids.To ensure the power grid voltage stability after the large-disturbance fault,a response-driven emergency control strategy is proposed based on convolutional neural network(CNN)and long short-term memory(LSTM)network.First,the mapping relationship between voltage time sequence value of key bus nodes and the voltage stability level is analyzed,and the large-disturbance voltage stability evaluation model based on CNN-LSTM network is established offline.Secondly,the evaluation model is used to predict the enhancement of power grid voltage stability level after the operation of alternative generator tripping and load shedding spot control measures and to determine the sensitivity of response-driven emergency control measures.Finally,considering the sensitivity of emergency control measures,an emergency control optimization problem is established with the actual operation constraints of the power grid,and the optimal emergency control strategy is obtained.For the actual scenario of hybrid AC/DC power grid with the voltage instability problem,the simulation results verify the accuracy of the proposed response-driven control sensitivity prediction method.The optimal coordination strategy of emergency control measures for voltage stability can ensure the safe and stable operation of power grid after large-disturbance faults.