This paper proposes a novel sensorless nonlinear controller for Photovoltaic Energy Conversion Systems (PECS) operating under stochastic climatic variations, uncertain system parameters, and plant-model mismatches. The inherent nonlinear dynamics of PECS, compounded by climatic variability and load & system perturbations, challenge system stability and maximum power point tracking (MPPT) efficiency. To address this, a Neuro-Climatic Sensorless Observer-based nonlinear controller (NCSOBN) has been introduced. To tackle the stochastic problem of the PECS which often imposes the need for expensive climatic sensing, a neural network estimates the reference voltage trajectory without costly climatic sensors. A robust Immersion & Invariance (I&I)-based controller then stabilizes the PECS by submerging its dynamics into a lower-dimensional invariant manifold, ensuring convergence to the desired trajectory and robust MPPT under climatic uncertainties, load variations, and parametric disturbances. Unmeasured variables, such as inductor current, are reconstructed via I&I observers, while a finite-time robust exact differentiator estimates voltage derivatives. Rigorous mathematical analysis substantiated by Lyapunov laws verifies the overall stability of the controller. From a control-theoretic perspective, benchmarks against PI, sliding mode, and integral Backstepping controllers demonstrate NCSOBN’s superior convergence speed (1.9 ms), zero steady-state error, and resilience to internal/external uncertainties. In the realm of PECS, benchmarking against existing MPPT methods, reveals surpassing efficiency exceeding 99.7 % with very rapid convergence. Embedded digital implementation and experimental verification using STM32 card confirms practical viability. This piece of work advances robust control paradigms for renewable energy systems, offering a computationally efficient, sensor-reduced solution for real-world PECS deployment.
Harrison et al. (Fri,) studied this question.