A Novel Method to Identify and Classify Deterioration of Orange Juice

The deterioration of orange juice poses significant health risks to consumers and undermines the integrity of the fruit juice market. Current detection methods, such as the Brix method, have limitations in specificity and practicality, often requiring sophisticated laboratory equipment and being time-consuming. This paper explores an innovative approach to detecting orange juice adulteration using microwave reflection techniques combined with advanced machine learning models. A Vector Network Analyzer (VNA) was employed to measure the dielectric properties of orange juice samples over a broad frequency range. Data were processed using advanced signal processing algorithms and various normalization techniques. Several machine learning models were evaluated, with the Cubic Support Vector Machine (SVM) achieving a final validation accuracy of 99.7%. This study presents a novel, non-destructive, and cost-effective solution for real-time detection of adulterants in orange juice, contributing to consumer safety and market integrity.