Neural Network-based State Feedback Control of Parallel UPS System

This paper proposes a state feedback control algorithm with estimation of the system parameters using neural network (NN) in parallel uninterruptible power supply (UPS) systems. Based on the state-space model of the parallel inverters, multi-input and multi-output (MIMO) control structure is used for tracking the AC voltage reference and minimizing the circulating current between submodules. In addition, the online parameters identification is achieved by using the neural network system. Then, the updating estimation results are adapted to improve the control performance. Extensive experimentation results of both software simulations and hardware experiments verify the validation and effectiveness of the algorithm.