基于结构化神经网络的微电网二次调频控制OA

With the increasing complexity and uncertainty of microgrid systems,designing secondary frequency regulation control methods that combine both stability and output tracking capabilities is crucial for the safe and reliable operation of the system.To this end,this paper proposes a secondary frequency regulation method for microgrids based on structured neural networks.First,by leveraging the equilibrium point independent passivity widely present in physical systems,the secondary frequency regulation problem for microgrids is formulated as a strictly monotonic proportional-integral(PI)control structure,and the PI parameters are parameterized as the gradient of a strictly convex neural network(SCNN).Second,the SCNN is constructed using a Softplus-β activation function with tunable coefficients,which not only ensures its universal approximation capability but also effectively adapts to the diverse communication constraints in microgrids.Furthermore,by employing the gradient of the strictly convex function constructed by the SCNN as a Lyapunov function,an end-to-end stability proof for the system is provided.Simulation results show that,compared with other unstructured neural network control methods which may lead to system instability,the proposed method not only guarantees system stability but also outperforms traditional control methods in both transient and steady-state performance.