Machine Learning based Cooperative Control of Photovoltaic Systems for Voltage Regulation

Penetration of photovoltaic (PV) sources is increasing in the distribution network (DN) which poses a concern of voltage rise due to a higher R/X ratio of DN. To tackle this, PVinverters connected to DN should take part in ancillary services. The objective of this work is to compensate the voltage rise using reactive power control. As many PV inverters need to contribute in reactive power control, the cooperative control mechanism is required. For the effective and reliable cooperative control, the communication among the various PV units is required which lead to increased cost and increased control complexity. Therefore, in this work, machine learning based i.e., neural network (NN)- based cooperative control strategy is established for the voltage regulation through reactive power control of PV inverters. The proposed method does not need costly communication infrastructure, needs only local information and it is still accurate. Moreover, this approach does not need droop control which has serious drawbacks. Through the simulations results, it is established that proposed method is highly effective for voltage regulation at various buses in DN.