•  
  •  
 

Abstract

This research provides an analytical study of the effect of fault resistance on the accuracy of fault classification in solar farms, with the aim of evaluating the performance of classification algorithms in realistic environments where the characteristics of electrical signals resulting from faults change. A model of a 290kW photovoltaic farm, consisting of 100 strings each containing 7 solar panels, was adopted, and the simulation was carried out using MATLAB/Simulink software. The study included different scenarios for electrical faults, such as line-to-ground (LG) and line-to-line (LL) failures, with the failure impedance changing from 0 to 500 ohms by a step of 20 ohms. Basic electrical signals such as currents, voltages, power, temperature, and radiation, were recorded for each fault. The data was pre-processed, then a classification model was built using the KNIME platform with the Artificial Neural Network (ANN) algorithm, and the data was divided into training and testing groups. The evaluation was conducted using multiple performance indicators: accuracy, retrieval, predictive accuracy, F1-Score, Matthews Correlation Coefficient (MCC), Area Under the Curve (AUC), and Logarithmic Loss (Log Loss), as well as confusion matrices analysis. The results showed that the high fault impedance leads to a gradual decrease in the performance of the model, as a result of the decrease in the clarity of the electrical signals associated with the fault.

Download

Included in

Engineering Commons

Share

COinS