基于回声状态网络控制的并网型太阳能光伏系统电能质量提升

This work presents an echo-state network (ESN)-based technique to enhance the utility power quality (PQ) performance in a grid-integrated solar photovoltaic (PV) system. In practice, the problem of PQ has become an issue of concern due to unprecedented utilization of highly nonlinear loads. This research presents an ESN-based control approach that performs the dual functionality of active power management and PQ improvement. This system supports current harmonics suppression, load balancing, neutral current elimination, grid reactive power reduction, and power factor improvement. An ESN approach extracts the harmonics-contaminated load currents' active and non-active harmonic-free fundamental weight components. This scheme delivers promising performance in terms of robustness and adaptation in a highly nonlinear load environment. The ESN-based control approach practically enables the online learning of weights of the load current harmonics to guarantee unconditionally balanced sinusoidal grid current. The effectiveness of the proposed ESN approach is validated through simulation. The practical viability of the proposed control has been demonstrated using the dSPACE 1104 real-time controller. The employed ESN-based control algorithm features faster convergence speed and low steady-state errors as compared to its competitive algorithms in time-variant scenarios. Extensive simulation and test results are demonstrated at various conditions while feeding nonlinear loads such as, load unbalancing and changes in solar insolation level. The experimental findings validate sinusoidal grid currents with enhanced PQ and unconditionally balanced grid currents. to comply with IEEE std-519 for the harmonic current limit.