Neural-network Estimation Coordinated State Feedback Control for Parallel Inverters Considering Filter Parameter Variation

A coordinated state feedback control scheme for the parallel-inverter system is presented in this paper. The proposed algorithm combines dynamic estimation and online parameters identification. It is intended to improve performance on both AC bus voltage reference tracking and circulating currents minimizing, especially during parameter variations caused by aging leading drifts and different environmental conditions. A neural network (NN) based dynamics estimation compensates for unknown practical dynamics and reconstructs the system. Based on this, a Lyapunov-function-based parameter identifier is proposed to achieve online parameter identification. Identification results are then adapted as feedback to the NN estimation model to improve steady and dynamic performance. A control framework is finally formed in a coordinated way. Simulation and experimental results indicate the algorithm can achieve high performance and identify time-varying line parameters in real-time to enhance system reliability.