Fruit Juice Dielectric Properties Characterization using Artificial Intelligence

This paper presents an innovative method for measuring the quality of fruit juice using Artificial Neural Network (ANN). An open-ended probe connected to a vector network analyzer (VNA) was used to measure the dielectric constant and loss factor. Both dielectric constant and loss factor values of these samples were analyzed over a frequency range from 1GHz to 6GHz. The project focuses solely on oranges, with 7 distinct varieties of oranges selected and prepared for analysis. By referring to the difference in pH, sugar content, and dielectric properties, the behavior of the dielectric constant and loss factor are analyzed. To develop the ANN model, 707 data points are used as input and target data. To achieve target accuracy, the model uses a data set with unbiased ANN design parameters, including a feed-forward backpropagation network type, Tan-sigmoid transfer function, 5 neurons in the hidden layer, and Levenberg-Marquardt backpropagation as the training algorithm. The model's performance is evaluated by comparing measured and computed values to ensure precision in predicting juice quality through MATLAB software.