Equality‐embedded augmented Lagrangian neural network for DC optimal power flow

Jiayu Han,Chao Yang,Lei Yan,Mengyang Niu,Yupeng Zhang,Cheng Yang

DOI: https://doi.org/10.1049/rpg2.13048

IF: 3.034

2024-07-22

IET Renewable Power Generation

Abstract:The proposed augmented Lagrangian neural network (ALNN) consists of two neural networks: the control net and the penalty net. The control net is used to mimic augmented Lagrangian function and predict control variables: active power of generators; the penalty net is used to update the Lagrangian multipliers. In this way, ALNN uses the augmented Lagrangian method to solve DC optimal power flow problem by two neural networks. Direct current optimal power flow (DC‐OPF) problems need to be solved more frequently to maintain safety and economic power system operation. Traditional solvers take too much time to get optimal results. To overcome it, a new self‐supervised augmented Lagrangian neural network (ALNN) is proposed to solve DC‐OPF problem. The proposed ALNN consists of two neural networks: the control net and the penalty net. The control net predicts active power of generators; the penalty net updates the Lagrangian multipliers. The equality constraints are embedded into the control net to guarantee no equality violations. The generalized reduced gradient method is used to reduce theviolations of inequality constraint. The effectiveness of the proposed model is demonstrated on IEEE 118‐bus. The results show that with the help of equality embedding, the equality constraints are always satisfied, which in turn improves the feasibility of ALNN. Compared to the state‐of‐art models, the proposed model has higher feasibility and less constraint violations without comprising optimality. What is more, most of the inactive constraints can be found during the training process and then they are used to speed up the post‐processing part.

energy & fuels,engineering, electrical & electronic,green & sustainable science & technology