Control of UPFC using Hamilton-Jacobi-Bellman formulation based neural network

In this paper, the micro grid stability is investigated by utilizing a non-linear optimal controller and FACTS device. Using micro grid continuous-time model and control design impose a huge computational burden due to the required high sampling rate to achieve stability when utilizing a digital controller. Thus, developing of an advanced discrete-time (DT) stabilizing controller design is of paramount importance in the micro grids. In this paper a nonlinear discrete-time stabilizing controller using Unified Power Flow Controller (UPFC) is proposed for micro grids by employing the discrete-time Hamilton-Jacobi-Bellman (HJB) optimal control method. The designed optimal controller is applied to control the UPFC´s series voltage and to optimally mitigate the power oscillations. The micro grid under consideration is comprised of a synchronous generator, renewable energy sources, and loads. The UPFC series voltage is considered as control input and the optimal strategy is applied. A discretized micro grid nonlinear dynamical model is derived and successive approximation method is utilized to approximate the cost function of the generator states and the UPFC control parameters. Finally, a neural network (NN) is utilized to approximate the cost function using the weighted residual method. By applying the developed optimal controller, it is shown that oscillations caused by faults are mitigated more effectively compared to the conventional generator controllers.