| Summary: | This paper proposes an end-to-end (<inline-formula> <tex-math notation="LaTeX">$E2E$ </tex-math></inline-formula>) learning-based control policy to directly control a transformerless grid-tied three-level neutral-point-clamped (<italic>3L-NPC</italic>) inverter powered by a photovoltaic (<italic>PV</italic>) array. This <inline-formula> <tex-math notation="LaTeX">$E2E$ </tex-math></inline-formula> control policy is represented by an artificial neural network (<italic>ANN</italic>) and a time-delay neural network (<italic>TDNN</italic>), namely, <italic>ANN</italic> - and <italic>TDNN</italic> -based control policies, to properly estimate the <italic>optimal</italic> switching vector of the <italic>3L-NPC</italic>. With such learning-based control policy, there exists no need for deriving or understanding deeply the complex mathematical model of the <italic>3L-NPC</italic>, as the dynamics of both the system and the control scheme, as well as the cost function to be minimized, are learned via an end-to-end learning fashion that maps directly from the raw observations to the <italic>optimal</italic> switching states. This definitely eliminates the major barriers of the model-based control strategies (i.e., model predictive control (<italic>MPC</italic>)) such as (i) the need for an accurate system model, and (ii) the exponential increase in computational complexity. In order to train the two control policies, the conventional <italic>MPC</italic> is employed, as an expert, for acquiring a set of training data (i.e., input-output pairs) and, thereafter, for assessing our proposed control schemes. The proposed <italic>E2E</italic> control strategies are validated using MATLAB/Simulink software, where the impact of having different input features and training data are studied. With the proposed control policies, especially the <italic>TDNN</italic> -based control policy that has only one time-delay window, a high-quality sinusoidal grid current is achieved with low total harmonic distortion (<italic>THD</italic>), resulting in enhancing the power quality of the utility grid. In addition, the leakage current is minimized compared to the conventional <italic>MPC</italic> by more than 25%. However, the same dynamic behavior is almost obtained during the irradiation changes compared to the <italic>MPC</italic> strategy. Moreover, the experimental verification of the proposed <italic>E2E</italic> control strategy is implemented on the basis of the Hardware-in-the-Loop (HIL) simulator using the C2000TM-microcontroller-LaunchPadXL TMS320F28379D kit, demonstrating the applicability and good performance of our proposed control strategy under realistic conditions.
|
|---|