| 초록 |
Plastic compounds have been developed by many experts having experience in various industries. However, it takes a lot of effort and time to acquire the properties of plastic compounds due to compounding mixing, injection condition and measurement method. We studied the properties prediction technology of plastic compounds by data mining and machine learning algorithms. To collect data on the compositions, processing and properties of plastic compounds, we built templates and database systems. Machine learning algorithms were applied to predict tensile strength, flexural modulus, Izod impact strength, and heat deflection temperature of plastic compounds. To improve the prediction performance, we adopted the classification and properties of materials used in plastic compounds. In addition, the clustering of similar fillers was visualized by applying principal component analysis method using the characteristics of the various fillers used in plastic compounds. |
