Design of a Power Flow Convergence Discriminator Based on Physical Information Model

As the structure of power systems becomes increasingly complex, the number of possible operating modes in power systems increases, leading to a sharp rise in the difficulty and computational load of traditional power flow calculations. Meanwhile, the application of artificial intelligence technology in new power systems still lacks rationality, making it challenging to apply in real grid operations. This paper proposes a power flow convergence discriminator based on a physical information model, utilizing a multi-Layer Perceptron network to implement a voltage predictor that satisfies the equality constraints of power flow calculations. Through power flow networks and residual calculations, it meets the inequality constraints to ensure the feasibility of actual operations. Simulations were conducted on the IEEE 30-bus system, and the results indicate that the power flow discriminator can accurately distinguish five types of power flow samples, achieving a testing accuracy of $96 \%$.