Edmund Coersmeier

Task9 GmbH, Germany

Title: Sensor Data Fusion for Wastepaper Sorting Process Analysis

---

Biography

Edmund Coersmeier completed his phd thesis at the age of 35 years at the Technical University Dortmund, Germany and became a professor for Applied Computer Science at the University of Applied Science Bochum, Germany 3 years afterwards. Beforehand he worked for 12 years at Nokia Research Center Helsinki and Bochum as Senior Research Engineer and Principal Scientist. Edmund is co-founder of the start-up company Task9 GmbH established in 2008 and acts as CTO with a clear focus on agent-based algorithm and system design, especially for complex closed-loop control and optimization tasks.

Abstract

Agent-based wastepaper sorting analysis should improve sorting quality as well as deinking yield if the right conclusions are drawn from the measured data. Because the wastepaper sorting process includes different sorting process steps and many different machine parameters, it is a complex procedure for the operator team in charge to find potential improvements in real time if the sorting process shows anomaly. The work investigates, where to position different cameras along the paper sorting stages and explains exemplary the close relationship between sorting results and electrical energy use for compressed air generation. Thus the paper describes the system behavior in terms of key performance indicators like impurity, energy costs, yield and throughput. The investigation starts on the input material composition and shows the relationship between sorting quality and yield, depending on the compressed air availability. Thus the paper draws an overall architecture on sensors and sensor data fusion to initially analyze, when and why wastepaper sorting quality might be not as good as expected. The key idea is to combine agent based camera images and energy data management system to fuse data from different sensor domains. The data is visualized within different time series plots as well as bar charts to describe system component relationship numerically as well as qualitative. Finally the work gives an outlook how sensors, algorithms and actuators should be combined to come to an automatic feedback control loop for guaranteeing a constant wastepaper sorting quality as well as deinking yield.