A Fast Dynamic Internal Predictive Power Scheduling Approach for Power Management in Microgrids

This paper presents a Dynamic Internal Predictive Power Scheduling (DIPPS) approach for optimizing power management in microgrids, particularly focusingon external power exchanges among diverse prosumers. DIPPS utilizes a dynamic objective function with a time-varying binary parameter to control the timing of power transfers to the external grid, facilitated by efficient usage of energy storage for surplus renewable power. The microgrid power scheduling problem is modeled as a mixed-integer nonlinear programmig (MINLP-PS) and subsequently transformed into a mixed-integer linear programming (MILP-PS) optimization through McCormick’s relaxation to reduce the computational complexity. A predictive window with 6 data points is solved at an average of 0.92s, a 97.6% improvement over the 38.27s required for the MINLP-PS formulation, implying the numerical feasibility of the DIPPS approach for real-time implementation. Finally, the approach is validated against a static objective using real-world load data across three case studies with different time-varying parameters, demonstrationg the ability of DIPPS to optimize power exchanges and efficiently utilize distributed resources whie shifting the eexternal power transfers to specified time durations.

💡 Research Summary

The paper introduces a novel Dynamic Internal Predictive Power Scheduling (DIPPS) framework designed to optimize power management in microgrids with heterogeneous prosumers, focusing particularly on the timing of power exchanges with the external grid. The core innovation lies in embedding a time‑varying binary parameter bS(t) into the objective function, which allows the scheduler to dynamically switch between two modes: (1) suppressing power sales to the grid and storing surplus renewable generation in the Energy Storage System (ESS) during designated periods, and (2) forcing the sale of stored energy regardless of the prevailing time‑of‑use (TOU) tariff during other periods. This dynamic objective enables the microgrid to align purchases with low‑price intervals and sales with high‑price intervals, thereby reducing overall electricity costs and improving the utilization of distributed resources.

The underlying optimization problem is first formulated as a Mixed‑Integer Non‑Linear Programming problem with Predictive Scheduling (MINLP‑PS). Non‑linearities arise mainly from products of binary decision variables and continuous power variables (e.g., charging/discharging decisions). To achieve real‑time solvability, the authors apply McCormick’s relaxation to linearize these bilinear terms, introducing auxiliary variables (z, y, w) that replace the products. The resulting problem, a Mixed‑Integer Linear Programming with Predictive Scheduling (MILP‑PS), retains all operational constraints: power balance, PV generation allocation, ESS state‑of‑charge dynamics, charging/discharging power limits, and mutual exclusivity of charging, discharging, and grid transactions enforced by three binary variables (bV, bG, bC).

Computational experiments use real‑world data: the UCI Energy Dataset for household load profiles (France, 2006‑2010), PV generation profiles, and TOU price signals. Simulations are executed in MATLAB R2020b on an Intel i7 machine with 16 GB RAM. The MINLP‑PS version solved with IPOPT requires an average of 38.27 seconds per 6‑step prediction horizon, whereas the MILP‑PS version solves in 0.92 seconds—a 97.6 % reduction, demonstrating feasibility for online operation.

Three case studies evaluate the impact of the dynamic objective. All cases share identical load, PV, and ESS capacities. Case A employs a static cost objective, leading the scheduler to minimize grid purchases; ESS charges only when PV surplus exists and discharges to meet load during low‑generation periods, with any remaining stored energy sold during the high‑price window (13–19 h). Cases B and C activate the binary parameter bS(t)=1 during 0–6 h and 18–24 h, respectively, forcing the microgrid to sell power in those intervals regardless of price. Results show that the timing of external power sales shifts according to bS(t), allowing the system to store cheap energy and release it when tariffs are favorable, thereby achieving lower total cost compared with the static case.

Key contributions of the work are: (1) a systematic transformation of a MINLP microgrid scheduling problem into a tractable MILP using McCormick relaxation, enabling sub‑second solution times suitable for real‑time control; (2) the introduction of a time‑varying binary parameter that provides flexible control over external power exchange timing, directly addressing TOU tariff structures; (3) validation with real‑world datasets across multiple scenarios, confirming that DIPPS improves cost efficiency and ESS utilization while respecting operational constraints such as state‑of‑charge limits and charging power caps.

The authors suggest future extensions including integration of battery degradation models, expansion to multi‑microgrid peer‑to‑peer trading frameworks, and development of distributed optimization algorithms to handle larger communities while preserving the dynamic objective’s benefits.