Real-time analytics to determine the quality of input in waste pre-treatment plants

In the framework of a larger project (ReWaste4.0), research work is carried out to char-acterise input into waste pre-treatment plants by means of real-time analysis. Commer-cial waste was selected as input material for the experiments. A number of samples were pre-processed (shredding, sieving) and sorted into a number of fractions. The main experiments will be carried out with individual waste objects, which are taken from nine sorting fractions. Regarding the real-time analysis, two approaches will be used, i.e. sensor-bases analysis (NIR-sensor/RGB-camera) and a deep learning ap-proach (image classification system). The produced data are related to data, which are generated by manual measuring (object size, weight) and laboratory analysis (heating value, water and chlorine content). By using regression analysis, the data of the real-time analysis are related to the data of standard laboratory analysis.