MODELING AND INVESTIGATION OF A SUPERVISORY ENERGY MANAGEMENT SYSTEM FOR A HYBRID POWER PLANT WITH A BATTERY SUBSYSTEM

Abstract

This paper proposes and examines the architecture of a supervisory energy management and monitoring system for a hybrid power plant that combines a photovoltaic subsystem, wind generation, and battery energy storage. The motivation is broader than dispatch tracking alone. The system must also limit supply deficit, reduce renewable curtailment, and restrain battery cycling under variable generation and delayed information channels. The study therefore aims to build a structured system architecture and to verify, in discrete time, the operability of a reserve-aware supervisory energy management algorithm that explicitly accounts for time synchronization, measurement filtering, sensor and command delays, as well as battery SOC and power constraints. The novelty lies in coupling a supervisory SCADA/EMS architecture with reserve-aware rule-based logic and assessing it by a joint set of energy, accuracy, and battery-usage KPIs in nominal, scenario-based, and stochastically perturbed modes. The object of the study is the distribution of energy flows in a hybrid plant with storage; the subject is reserve-aware rule-based supervisory control that preserves battery charge for evening and night intervals. The methodology combines mathematical modeling, scenario analysis, a 24-hour simulation with a 60 s supervisory step, 30 Monte Carlo realizations, and comparison with a reference rule-based controller without reserve, a naive bang-bang controller, a no-battery case, and an additional offline optimization baseline. In the baseline case with a 240 kWh battery and a 60 kW power limit, the proposed controller achieved 68.74 % schedule energy tracking, a deficit of 701.88 kWh, curtailment of 1.74 kWh, RMSE of 38.91 kW, and 0.387 equivalent deep cycles. Relative to the reference controller without reserve preservation, it lowers RMSE from 40.83 to 38.91 kW and reduces deep-cycle equivalent from 0.401 to 0.387, but increases the deficit by 6.12 kWh and decreases K_track by 0.27 percentage points. This reflects a trade-off between tracking accuracy, battery preservation, and reserve retention. Raising battery capacity to 400 kWh cuts the deficit to 661.38 kWh, whereas the high-generation scenario increases schedule tracking to 78.45 %. Across 30 Monte Carlo realizations, the proposed controller yielded mean values of E_def = 783.84 ± 153.42 kWh, RMSE = 42.53 ± 5.99 kW, K_track = 65.73 ± 6.51 %, and N_cycle,eq = 0.292 ± 0.148; descriptive paired comparison with the reference controller showed lower RMSE and lower cycling in 100 % of realizations, although with a slightly higher energy deficit. The additional offline optimization baseline, supplied with full knowledge of the daily profile, confirmed that the conclusions are sensitive to the chosen objective function and should not be read as a universal upper quality bound. The maximum control-cycle computation time did not exceed 1.48 ms in the simulation environment, which indicates low computational complexity for a 60 s supervisory cycle. The results should still be interpreted as simulation- based because the study relied on representative rather than measured or archival time series.