Design of an MPPT Algorithm Based on Neural Networks and Fuzzy Logic for Electromobility Applications

Rivera W.; Fernandez D.; García M.; Angulo J.; Töfflinger J.A.; Paragua-Macuri C. · Lecture Notes in Networks and Systems (2026) · DOI: 10.1007/978-3-032-18982-0_34

Resumen

Electromobility platform solar-assisted EVs and rail cars experience abrupt irradiance swings and frequent partial shading that depress PV energy yield and destabilize conventional MPPT schemes. We propose a hybrid MPPT that couples a rule-based fuzzy-logic controller with a lightweight artificial neural network (FL-ANN) to enhance tracking speed, accuracy, and robustness. A DC–DC boost converter enacts duty-cycle commands; in parallel, technology-specific ANNs (HIT, a-Si/c-Si tandem, a-Si, and CIGS) estimate MPP voltage and power. Each ANN is trained on more than five thousand field/logged samples per technology collected between Jan 2024 and Mar 2025. MATLAB/Simulink studies under Standard Test Conditions and dynamic scenarios irradiance steps, 25–40 ∘C temperature sweeps, mobility-inspired partial shading, and sensor noise show consistent gains over standalone fuzzy logic: average tracking efficiency rises 98.71% at STC and 99.12% at 40 ∘C; settling times shorten by 12–25%, and output ripple decreases across all tests. The FL-ANN-assisted controller has medium implementation complexity on low-cost microcontrollers, requires modest memory, and generalizes across four PV technologies without retuning. These attributes make it an implementation-ready solution aimed at maximizing energy capture in mobility-oriented PV systems. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

Instituciones: Pontificia Universidad Católica del Perú