Optimization of active transmission network partitioning for high-risk line cascading failure prevention

Electrical distance is widely used as edge weights in spectral clustering for transmission network partitioning. However, the propagation risk of cascading failures is seldom embedded into the graph model. The emergency support potential of flexible resources also remains underutilized. As a result, power imbalance and excessive load shedding may occur in sub-networks after partitioning. In this paper, an active partitioning optimization method is proposed for transmission networks to prevent cascading failures in high-risk lines. A line fault propagation interaction matrix is built through Monte Carlo simulation, and a propagation risk index is defined accordingly. The propagation risk index is combined with line loading rate to form a comprehensive risk metric. The comprehensive risk metric is mapped onto graph edge weights via a Gaussian kernel function. An initial partitioning scheme is then generated by spectral clustering. A cut-set screening strategy based on source-load balance constraints is further developed. Heuristic boundary correction and electric vehicle aggregator (EVA) emergency support are introduced to guarantee power equilibrium in each sub-network. The method is tested on the IEEE 39-bus system under high-risk line fault scenarios. Compared with the natural evolution of cascading failures, power balance is achieved in all sub-networks after active partitioning, and load shedding is significantly reduced. Multi-strategy coordination is shown to be effective in preventing cascading failure propagation.