Prediction of Acrylonitrile–Butadiene–Styrene\nMechanical Properties through Compressed-Sensing Techniques

One of the challenges\nin the plastic industry is the cost and time\nspent on the characterization of different grades of polymers. Compressed\nsensing is a data reconstruction method that combines linear algebra\nwith optimization schemes to retrieve a signal from a limited set\nof measurements of that signal. Using a data set of signal examples,\na tailored basis can be constructed, allowing for the optimization\nof the measurements that should be conducted to provide the highest\nand most robust signal reconstruction accuracy. In this work, compressed\nsensing was used to predict the values of numerous properties based\non measurements for a small subset of those properties. A data set\nof 21 fully characterized acrylonitrile–butadiene–styrene\nsamples was used to construct a tailored basis to determine the minimal\nsubset of properties to measure to achieve high reconstruction accuracy\nfor the remaining nonmeasured properties. The analysis showed that\nusing only six measured properties, an average reconstruction error\nof less than 5% can be achieved. In addition, by increasing the number\nof measured properties to nine, an average error of less than 3% was\nachieved. Compressed sensing enables experts in academia and industry\nto substantially reduce the number of properties that must be measured\nto fully and accurately characterize plastics, ultimately saving both\ncosts and time. In future work, the method should be expanded to optimize\nnot only individual properties but also entire tests used to simultaneously\nmeasure multiple properties. Furthermore, this approach can also be\napplied to recycled materials, of which the properties are more difficult\nto predict.