Data-Driven Takagi–Sugeno Fuzzy Modelling for Photovoltaic Maximum Power Point Using an Improved Grey Wolf Optimizer

The rapid expansion of photovoltaic (PV) deployment has increased the demand for accurate and interpretable models for forecasting, controlling, and monitoring conditions. We present a data-driven identification procedure that learns Takagi–Sugeno (T–S) fuzzy rule bases directly from plant measurements by using an Improved Grey Wolf Optimizer (I-GWO). Surrogate models predict the current and voltage at the maximum power point (MPP) as functions of the irradiance and module temperature. The validation of a 9.54 kW grid-connected installation in Algiers, Algeria, demonstrated that the learned rule bases were compact with local linear consequents fitted to time-aligned meteorological and electrical data. Under clear and cloudy sky conditions, the models reproduced the measured MPP quantities with root-mean-square (RMS) errors of current 0.1046 A and voltage 19.64 V, corresponding to a power error of 19.21 W. Under partly cloudy conditions, the fidelity remained high, with RMS errors of 0.5265 A and 45.43 V and a power error of 182 W. These results indicate that the approach achieves a practical balance between predictive performance and transparency while obviating equivalent circuit parameter identification. The framework is broadly applicable and can be extended to other renewable energy assets that require robust and interpretable surrogate models.