THE TWO-LAYER ENERGY MANAGEMENT AND DISPATCH MODEL FOR MICROGRID HYBRID ENERGY STORAGE SYSTEMS BASED ON MPC

Abstract

As the penetration of distributed photovoltaic (PV) and wind generation in microgrids continues to rise, the strong stochasticity of renewable output and load poses significant challenges to the economic operation of microgrids, while the lifetime degradation of energy storage batteries under frequent charging and discharging further inflates the whole-life-cycle cost of the system. To address this issue, this paper proposes a two-layer energy management and dispatch model for microgrid hybrid energy storage systems (HESS) based on model predictive control (MPC). The upper layer is an economic dispatch layer, which minimizes the sum of the electricity purchase cost and the battery degradation cost and adopts a 48-step receding-horizon optimization to determine hourly reference plans for battery charging/discharging and grid interaction; the degradation cost is modeled as a depth-quadratic function of the half-cycle throughput and of the lifetime parameters, thereby internalizing lifetime loss into the economic objective. The lower layer is a power smoothing layer, which tracks the upper-layer reference over a 12-step short horizon and exploits the fast response of the supercapacitor to smooth the high-frequency power fluctuations induced by renewable forecasting errors. A 48-hour closed-loop simulation is carried out using real hourly data containing PV and wind generation together with a real-time price series. The results show that the upper-layer dispatch achieves peak shaving and valley filling by charging at low prices and discharging at high prices, with a correlation coefficient of −0.620 between the electricity price and the grid interaction power; the battery state of charge (SOC) operates strictly within the safe range of [10%, 90%] throughout, with zero violations. Compared with the baseline strategy that ignores degradation, the degradation-aware strategy reduces the cumulative battery throughput by 44.0%, the standard deviation of power fluctuations by 32.1%, and the total cost by 6.2%. At a 40% forecasting-error level, the lower-layer supercapacitor raises the reduction in the root-mean-square battery power ramp to 41.2%, and the smoothing benefit becomes more pronounced as the forecasting error increases. The proposed model provides a feasible hierarchical scheduling approach for the optimal operation of microgrids with HESS, balancing operational economy and storage lifetime under uncertainty.