Battery energy storage systems (BESSs) play a vital role in modern smart grids, yet their economic operation remains challenging. A fundamental problem lies in distributing load efficiently across multiple battery units while minimizing losses from internal impedance and managing capacity degradation—both of which significantly increase operating costs.
Traditional economic dispatch algorithms rely on marginal cost (MC) consensus using proportional (P) controllers. However, these approaches suffer from slow convergence speeds and limited control accuracy, making them impractical for real-time isolated grid operations.
To address these shortcomings, researchers have proposed a novel PI+R (proportional-integral with reset) control scheme designed specifically for multi-agent battery networks. The innovation lies in the reset mechanism: the integral term is strategically reset to zero when the proportional term crosses zero, effectively eliminating overshoot while accelerating convergence toward equilibrium. This hybrid approach combines distributed decision-making with centralized coordination during critical operational moments.
The mathematical framework ensures system stability by analyzing eigenvalues under the PI+R controller and introducing a dwell-time mechanism to prevent Zeno behavior—a phenomenon where the system makes infinite transitions in finite time. The scheme also incorporates handling for input delays, a practical consideration in real grid communications.
Critically, the researchers modified the MC scheme to respect battery capacity constraints, ensuring the controller cannot demand charging or discharging beyond physical limits. This prevents equipment damage and extends battery lifespan.
Simulation results demonstrate that the PI+R controller achieves faster load-balancing across battery units compared to conventional P controllers, with improved steady-state accuracy and minimal overshoot. This advancement has significant implications for isolated microgrids, island power systems, and remote installations where energy storage reliability and cost-efficiency are paramount. The distributed nature of the algorithm also enhances scalability and resilience to single-point failures.
