An enhanced buck-boost converter for photovoltaic diagnosis application: Accurate MPP tracker and I-V tracer

Abstract

This study introduces a novel dual-functioning buck-boost converter designed for fault detection and diagnosis in photovoltaic (PV) arrays. The adopted control and diagnosis approach enables the converter to operate in two distinct modes depending on the state of the array. In normal operation, the converter is controlled by neural network (NN) controller to efficiently extract maximum power point (MPP). However, in the event of a system failure, the converter automatically transitions to variable load mode to capture different points on the current-voltage (I-V) curve. The transition between the two operational modes is ensured by a diagnosis system based on power loss analysis. For experimental purposes, a resistive load is employed as a simplified tool to characterize the system behavior and evaluate the performance of the converter in both operations. Experimental results confirm the functionality and accuracy of the proposed system, achieving high maximum power point tracking (MPPT) values of 0.59 % for MAPE and 0.993 regression compared to reference power. This precision contributes to improving the diagnosis program judgement to initiate the characteristic tracing. Furthermore, the system exhibits accurate tracing capabilities, with an average error of 1.44 % in case of normal operation. Similar errors are maintained even under diverse fault conditions, ranging from 0.77 % to 1.83 % for different faults including short-circuit, shunted panels, and connection faults. However, the error slightly increases in cases of partial shading fault, the effect and signature of fault remain clearly noticeable on the traced characteristics.