Cascading Failure Chains Screening Model of AC-DC hybrid Power System

With the operation of substantial HVDC lines, the risk of cascading failures is increased greatly. The propagation mechanism of cascading failures in AC/DC hybrid systems is analyzed, and a complete cascading failures chains search model for AC/DC systems in large-scale power grids is established. A method for rapid evaluation of AC faults to DC is proposed, which is divided into two steps. At first, the short circuit ratio of the system after the occurrence of the failure chain is calculated, and the extreme failure chain is identified. Time domain simulation is performed to whether it will trigger DC block. For the non-extreme accident chain, the magnitude of the fault is quantified by the area where the voltage after the fault exceeds the normal voltage value. The Leven berg-Marquardt neural network (LMNN) is used to quickly calculate the voltage dips degree, and the time domain simulation check is performed on the severe fault. The line failure probability index and the line failure risk index are defined. The pruning search is carried out according to the line outage risk after the occurrence of the accident chain at each level, which reduces the search space ensuring the search accuracy and greatly improves the search efficiency. Finally, the effectiveness of the method proposed to search high-risk cascading failures chains in AC-DC Hybrid Power System is verified by the simulations concerning the Shandong power grid of China.