In the petrochemical industry, steam reforming furnaces play a crucial role in large-scale hydrogen production. These furnaces are equipped with centrifugal cast HP-modified austenitic stainless-steel tubes, which are exposed to temperatures ranging from 600 to 1000ºC for extended periods. As a result, the wall temperature profile of these tubes exhibits a vertical gradient of distinct aging states, labelled as state I to VI in the literature, each characterized by specific microstructures. Given that the reformer tubes are expensive components of the furnace assembly, it is imperative to monitor their service life. Non-destructive testing is a vital tool for evaluating the structural integrity of industrial components. Hence, the objective of this study is to establish a real-time classification system for determining the aging states of HP-modified stainless-steel tubes using a non-destructive magnetic system and the machine learning technique Support Vector Machine (SVM). Two tubes, each measuring 12.6 meters in length, were removed from the same steam reforming furnace after 160,000 hours of service. The inspection was conducted using an eddy current hybrid probe adapted to an inspection vehicle, allowing for real-time data acquisition. The results demonstrated that the developed classification system was capable of accurately identifying the different aging states present along the studied tubes.