Coordinated Volt/VAR control of distribution networks based on mixed-integer intelligent optimization

Abstract

High penetration of photovoltaic generation causes simultaneous overvoltage, undervoltage, reverse power flow, and frequent operation of conventional regulation devices in active distribution networks. This paper proposes an adaptive mixed-integer differential evolution method with continuous polishing (AMIDE-CP) for coordinated Volt/VAR control. The method jointly schedules an on-load tap changer, stepped capacitor banks, and continuous reactive-power outputs of smart inverters. A mixed discrete-continuous chromosome is combined with integer repair, inverter-capability projection, a switching-cost model, exact radial AC power-flow evaluation, diversity-dependent mutation and crossover, and a final continuous polishing stage. The architecture can accept time-stamped measurements from fiber-optic voltage/current sensors and issue validated commands over optical Ethernet, thereby providing a practical automation-control pathway. A reproducible 24-h study on the IEEE 33-bus feeder compares the proposed method with a rule-based controller and a standard mixed-integer differential evolution baseline. AMIDE-CP reduces daily energy loss by 6.56%, root-mean-square voltage deviation by 63.80%, and mechanical switching operations by 60.0% relative to rule-based control. Relative to the standard optimizer, it reduces energy loss by 3.43%, root-mean-square voltage deviation by 3.59%, and switching operations by 50.0%, while maintaining every bus within 0.95-1.05 p.u. The average solution time is 0.438 s per control interval. Monte Carlo tests with forecast perturbations and one-period measurement staleness produce no voltage violations for AMIDE-CP, supporting its suitability for fast coordinated distribution-network automation.