Klasifikasi Tingkat Konsentrasi Larutan Berdasarkan Citra Pola Difraksi Menggunakan Metode Convolutional Neural Network (CNN)

Abstract

Determining solution concentration is a crucial parameter in various scientific fields, such as chemistry, physics, and industry. One method that can be used to identify solution concentration is through the analysis of light diffraction patterns formed by the interaction of light waves with the solution medium. Advances in image processing and artificial intelligence technology allow for automated diffraction pattern identification using deep learning methods. This study aims to implement and evaluate the performance of a Convolutional Neural Network (CNN) model in classifying solution concentration levels based on diffraction pattern images. This study used three types of solutions: sugar, sodium chloride (NaCl), and methylene blue, with varying concentration levels. The diffraction pattern images generated from each solution were used as the dataset to train a CNN model with the MobileNetV2 architecture. The model training process was carried out using the preprocessed image dataset. Model performance was then evaluated using accuracy, precision, recall, and F1-score metrics. The model was analyzed using a classification report, an accuracy-loss graph, and a confusion matrix. The results showed that the CNN model was able to classify solution concentration levels based on diffraction pattern images with quite good performance. The model achieved an accuracy of 95.73% for NaCl solution, 90.47% for methylene blue solution, and 80.00% for sugar solution. These results indicate that differences in the optical characteristics of each solution affect the resulting diffraction pattern, thus impacting the model's ability to extract image features. Overall, this study demonstrates that the CNN method based on diffraction pattern images has the potential to be used as an alternative approach in automatically identifying and classifying solution concentration levels.