A Hybrid GA–Digital Twin Strategy for Real-Time Nighttime Reactive Power Compensation in Utility-Scale PV Plants

This study proposes a hybrid method that combines a Genetic Algorithm (GA) with Digital Twin (DT) technology to address nighttime reactive power backfeed in large-scale photovoltaic (PV) power plants. First, the GA is employed to optimize the location and number of multitask inverters to minimize line losses and eliminate the reactive power backfeed. Subsequently, the DT continuously monitored the grid conditions and performed rolling dispatch to mitigate the residual reactive power caused by nighttime voltage fluctuations. Simulation results show that GA-based optimization reduces line losses from 0.346 to 0.2818 kW (18.6% reduction) and helps alleviate inverter thermal stress. When integrated with DTs, the method further improves voltage stability and demonstrates a strong adaptive control capability. The proposed GA–DT strategy can also be regarded as a potential AIoT application in PV plants, with the potential to reduce operational and maintenance costs and enhance the system reliability in the future.