A hybrid global maximum power point tracking of a PV-Battery system under complex partial shading conditions using horse herd optimization algorithm and a variable step size perturb and observe

Abstract

Photovoltaic (PV) arrays are sensitive to variable environmental conditions, which requires continuous regulation to maximize power, this makes maximum power point (MPP) trackers an essential element in every PV generation system, in particular, when PV arrays are under nonuniform irradiance caused by partial shading, the power voltage curve exhibits multiple maximums, which requires advanced MPP tracking algorithms to distinguish global MPP (GMPP) from local ones (LMPP), metaheuristic based MPPT algorithms were introduced for this particular reason, as they can catch GMPP under partial shading conditions (PSC). In this work, we proposed an improved metaheuristic algorithm by combining the horse herd optimization algorithm (HHOA) with a variable step-size perturb and observe (VSS-P&O) for MPPT of a PV-Battery system. The proposed HHOA-VSS-P&O solves the PSC problem and guarantees continuous tracking of the GMPP under variation in the optimal duty. The HHOA-VSS-P&O also reduces the search phase of the HHOA by reducing the number of iterations needed to reach the GMPP and consequently increasing the MPPT speed. Simulation results in MATLAB/Simulink, across four test scenarios, uniform step changes, standard PSC, multiple moving shadows, and combined edge-of-cloud/hotspot conditions- demonstrate the effectiveness of the proposed method against five other controllers (P&O, VSS-P&O, PSOA, GWOA, and HHOA). The proposed HHOA-VSS-P&O achieved the lowest tracking error among all tested methods in the two scenarios where quantitative tracking-error data were reported, with an RMSE of 9.70 W under the most severe multi-shadow condition and 14.2 W under combined edge-of-cloud and hotspot conditions. Under multiple moving shadows, the extraction efficiency was up to 99.3%, while edge-of-cloud and hotspot conditions yielded an efficiency of 97.7%, and the standard PSC yielded an efficiency of 96.4% with an average efficiency of 97.5% over all the tested scenarios. Regarding the steady-state power oscillation, the proposed method continuously outperformed the classical P&O and VSS-P&O (standard deviation of 28.6–43.2 W versus up to 89.6 W for P&O), but at the cost of a marginally higher oscillation than the pure metaheuristic controllers (PSOA, GWOA, HHOA), which was a conscious trade-off for its continuous-tracking exploitation phase. The proposed method also achieved reaching and settling times as low as 0.08 s and 0.10 s, respectively, with a typical number of iterations of 3, against incomplete or failed convergence for classical P&O and VSS-P&O under several shading zones. It was also observed that the computational time has been enormously reduced to 90 ms at moderate complexity of the proposed work as compared to the computational time of different works. This is significantly higher than the efficiencies of competing metaheuristic and classical controllers (including the classical controller) with efficiencies from 80.7% to 93.83% and processing times of up to 180 ms.