RANCANGAN APLIKASI DETEKSI DAN KLASIFIKASI KERUSAKAN PADA PANEL SURYA

Abstract

Solar panels are electronic devices designed to capture solar energy and convert it into electricity. Solar panels generally have a long lifespan and require little maintenance. However, like other devices, they can also experience damage or problems that impact the energy produced. To assist in the process of maintaining solar panels, an application is needed that can help detect and classify damage to solar panels. In this research, the method used to build an application interface for damage detection and classification on solar panels is to use a prototype model with the first stage starting with needs analysis, fast or simple design, building a prototype, evaluation, repair, evaluation and implementation. Meanwhile, to build and develop deep learning in this research, CRISP-DM (Cross-Industry Standard Process for Data Mining) is a systematic and structured approach used to manage data mining projects. By applying the Convolutional Neural Network Method to detect and classify damage to solar panels. In this research, the dataset used comes from the Kaggle platform, totaling 1442 images. The solar panel image is divided into 4 classes normal, hotspot, dirt, cracked and broken using Jupyter Anaconda as a tool and the level of accuracy will be measured using a confusion matrix. So it is hoped that the results obtained from this application will show a high level of accuracy with an interface that makes it easy for users.