SPARK: A Practical Framework for Solar PV Panel Diagnostics and Performance Evaluation

Abstract

Solar photovoltaic (PV) power plants play a critical role in clean energy generation, but their performance is often compromised by partial shading, dust, aging, and open-circuited bypass or antiparallel diodes. Timely detection of such faults is essential to prevent power losses, reduce downtime, and ensure safe operation. This paper presents SPARK (Solar Panel Analyzing and Recording Kit), a low-cost and practical framework for the real-time diagnostics and performance evaluation of solar PV panels. Unlike conventional approaches that require multiple current and voltage sensors, the proposed SPARK system employs a voltage-based diagnostic technique that significantly reduces sensor count and overall cost. A switching circuit with insulated gate bipolar transistors (IGBTs) and a minimal set of Hall-effect voltage sensors is designed to sequentially monitor the voltages of multiple panels in an array. The system integrates irradiance and temperature sensors, whose data, along with measured voltages, are processed by an ATmega-based controller running the proposed diagnostic algorithm. The algorithm estimates panel voltages under given conditions and compares them with measured values against adaptive thresholds to classify panels as normal, partially shaded, physically damaged, or faulty due to open-circuited antiparallel diodes. The hardware implementation is validated on a 2 kW rooftop PV plant consisting of six panels in two strings, with experimental testing under no-load and load conditions. Results confirm that SPARK accurately identifies normal and abnormal conditions, with LED indicators providing intuitive fault visualization. The proposed system offers a compact, scalable, and effective solution for condition monitoring of solar PV arrays in practical applications. Sensitivity and robustness analyses were conducted to ensure dependable fault detection across varying environmental and operational conditions.