Robust Optimal Control of Smart Grid Energy Distribution Under Demand Uncertainty

Abstract

The growing adoption of battery energy storage and renewable energy resources systems has created a unique requirement for energy management in smart grids. The performance of the conventional energy management strategies is considerably affected by electricity-price fluctuations, battery operating constraints, and short-term load forecasting errors. The battery energy management is illustrated through the proposed Adaptive Robust Model Predictive Control (ARMPC) framework under certain load demands. According to recent forecasting errors and short-term load variability, the receding-horizon model predictive control with an adaptive uncertainty set is combined, in which the size changes. The proposed framework explicitly accounts for forecast uncertainty, unlike deterministic MPC where optimisation takes place against a single forecast trajectory. The maintenance of the same conservative uncertainty margin is avoided under both variable and stable operating conditions, unlike fixed robust MPC. The models in the proposed framework involve battery degradation, grid exchange, photovoltaic generation, state of charge, battery charging and discharging, and load uncertainty within a rolling optimization problem. The planned numerical evaluation involves a 24-step prediction horizon, a 200-kWh battery, and a 200 kWh battery. The benchmark controllers involve fixed robust MPC and deterministic MPC. Peak grid import, operating cost, computational time, battery throughput, renewable-energy utilization and state-of-charge violations are used in the evaluation process. The proposed method indicates the following simulation values:  95.7% renewable-energy utilization, peak grid import of 82.4 kW, and an operating cost of 178.9 $/day. This study is anticipated to offer a practical foundation for embedding adaptive uncertainty handling into battery energy management, while keeping a clear balance among computational complexity, robustness, and operating economy.