Microstructural Classification of Unmodified and Strontium Modified Al–Si–Mg Casting Alloys with Machine Learning Techniques

抄録

The Al–Si–Mg casting alloy was modified by strontium, and the microstructural classification of unmodified and Sr modified samples was accomplished by using our originally developed methods and machine learning techniques. The classification rates of unmodified and modified samples were at high levels and the highest rate reached 97.5% accuracy when using statistical data and the support vector machine as the classifier. The additive of Sr caused the distribution of eutectic-Si particles to change from a random distribution to a clustering arrangement, and decreased the particle size of eutectic-Si. The tensile properties of the alloy were significantly increased after modification due to the refinement of the eutectic-Si phase.

Fig. 1 The flow diagram of the machine learning process in this work.