Abstract

The adoption of solar-powered water pumps is crucial for sustainable water management, particularly in remote and agricultural areas. However, the reliability of these systems is essential, as faults can result in significant downtime and expensive repairs. This paper introduces an intelligent fault detection model using logistic regression, a machine learning technique suitable for binary classification of faults. It leverages real-time sensor data, including electrical and mechanical parameters, alongside environmental conditions, to predict potential faults. The model was trained on a dataset from operational installations, undergoing pre processing for feature enhancement and missing data management. Evaluating the model involved key performance metrics such as accuracy, precision, recall, and the Area under the ROC curve, which confirmed its effectiveness in accurately detecting faults with high precision. A comparative analysis with decision trees and support vector machines emphasized logistic regression’s efficiency. The research contributes to intelligent monitoring systems and identifies future directions for enhancing fault detection methods.