Two-stage reactive power optimization for photovoltaic distribution networks based on improved whale algorithm

The prevalence of PV grid connections often leads to frequent voltage exceedances and increased network losses. To address this issue, we propose a two-stage reactive power optimization approach based on enhancements to the whale algorithm, leveraging the temporal characteristics of various reactive power compensation devices. Initially, predictive data is utilized to adjust discrete action devices at hourly intervals, aimed at minimizing network losses. Subsequently, intra-day real data is employed to regulate continuous devices at 15-minute intervals, considering both network losses and voltage stability. Moreover, we introduce an improved whale algorithm incorporating adaptive social learning and wavelet mutation mechanisms to address the optimization problem. This algorithm features an adaptive social learning strategy to foster population diversity and exploits wavelet functions to facilitate escaping local optima, thereby enhancing computational accuracy. Finally, we conduct simulation analyses using an enhanced IEEE 33-buses model to validate the efficacy of our proposed model and method.