DETECTING AND CLASSIFYING HARD AND SOFT INCLUSIONS IN LIQUID METAL

Apparatus and method are disclosed for providing a classification of inclusions in molten metal. The apparatus and method include the steps of and means for obtaining an analog signal stream from a data collection apparatus, passing the analog signal stream through an analog to digital converter to convert the analog signal stream to a digital signal stream, and partitioning the digital signal stream into a discrete time frame of 5 milliseconds. The digital signal stream is vector normalized to provide a normalized signal stream having a largest amplitude magnitude of 1. The normalized signal stream is compared to a control of a prototype shape for determining a classification for the digital signal stream over the discrete time frame by decision logic to determine hard inclusions versus soft inclusions in the liquid metal. In one aspect, the normalized signal stream is passed through a decision module having at least one threshold unit and at least one digital logic table to make a decision of a soft, deformable inclusion or a hard inclusion. In one embodiment, the classification and size of the inclusions are counted in an histogram of classification and a histogram of size. The classification and size of the inclusions are viewed in a graphical user interface.