Research on Two-Stage Coordinated Optimal Control of PV-ESS Voltage in Active PV Distribution Network Based on Inverter Residual Capacity

Abstract

Under high-penetration photovoltaic integration, node voltage over-limits are frequently experienced by active distribution networks during light-load periods at noon due to reverse active power flow, while voltage sag risks are faced during nighttime or sudden drops in photovoltaic output. As a result, voltage fluctuations exhibit bidirectional violation characteristics. Existing countermeasures tend to be polarized: fast control relying on local measurements, though timely in response, is prone to voltage deviation accumulation and regional circulating currents due to the lack of global coordination; global optimization based on iterative communication, though economically superior, is hampered by a convergence speed that can hardly match the second-level fluctuations of photovoltaic output. Moreover, by neither approach can the active and reactive remaining capacities of inverters be dynamically called upon according to their real-time operating states, and thus the conflicts among voltage regulation speed, accuracy, and economy are made difficult to resolve. Therefore, this paper proposes a two-stage photovoltaic active distribution network photovoltaic storage voltage collaborative control optimization method based on the remaining capacity of the inverter. By analyzing the coupling effects of photovoltaic-storage active power output on node voltages and network losses, the dominant role of active power regulation in distribution-network loss reduction is revealed. Dynamic safety factors and harmonic correction factors are introduced to quantify the effective remaining capacity of inverters in real time, so that the regulation boundaries are adaptively updated with operating states. Based on local measurements, dynamic droop control is implemented to achieve millisecond-level autonomous suppression of voltage violations. Furthermore, a regional coordinated optimization model is constructed with the objectives of minimizing network losses and voltage deviations. Local commands are corrected through limited information exchange among adjacent regions, and thus voltage deviation accumulation and regional circulating currents are eliminated at very light communication cost. The test results show that this method can determine the actual active power command correction and reactive power command correction at different over-limit depths. After control, the reactive power output of the photovoltaic inverter fluctuates less and reduces network loss by over 40%. The global voltage throughout the day is stabilized within a narrow range of 0.95–1.01 p.u., verifying the comprehensive advantages of this method in terms of fast and slow time scale coordination and dynamic capacity utilization.