Evaluation of MPC and ELM-Based Control Strategies for Maximum Power Point Tracking in Grid-Connected Photovoltaic Systems

Engineering notes

Engineering notes will be added by the Power for Solar editorial team.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为光伏效率、钙钛矿太阳能电池、储能技术、太阳能热利用、BIPV、并网技术等高价值论文补充中文说明。

Original abstract

Abstract This work benchmarks three maximum power point tracking (MPPT) strategies for photovoltaic (PV) systems, model predictive control (MPC), an extreme learning machine (ELM), and the classical perturb-and-observe (P&O) approach, under realistic, year-long operating conditions. We run time-series simulations driven by NASA POWER meteorological records for a tropical site in Vietnam (10.95°N, 108.10°E) on a 4.8 kWp PV array interfaced with a boost converter, and we quantify performance using annual energy yield, MPPT efficiency, transient tracking time, and steady-state power ripple. Across the full year, ELM delivers the largest energy yield (6377.5 kWh) and the highest mean MPPT efficiency (99.39%), while maintaining low ripple (0.37%), which suggests strong steady-state regulation under slowly varying irradiance. MPC reaches the fastest dynamic behavior, with a mean tracking time of 6.0 ms, markedly shorter than ELM (45.1 ms) and P&O (87.6 ms), but its annual energy yield (6352.4 kWh) remains slightly below ELM, consistent with a small trade-off between aggressiveness in transients and steady-state extraction. P&O records 6325.2 kWh and 98.88% efficiency, corresponding to a 0.82% annual energy shortfall relative to ELM in this dataset. Paired t-tests indicate statistically significant differences among all method pairs (p < 0.05). Overall, ELM appears preferable when long-term energy harvesting and low ripple dominate the design objective, whereas MPC aligns better with applications that prioritize rapid response to abrupt irradiance changes in grid-connected or standalone PV installations.

Links and sources

• PDF from original source

Comments