Autotuning of a vacuum conveyor

Abstract: The modern industry is steadily facing new demands for more energy efficient solutions. These improvements can come from technological advances, using existing technology smarter or a combination of the two. Today a vacuum conveying system is manually tuned for each factory application when installed and is in most cases not handled afterwards except for maintenance reasons. This in combination with the difficulties in tuning a system to run efficiently and robustly whilst offering high capacity indicates that a large number of systems are running with non optimal settings. The aim of this thesis was therefore to explore the possibility of introducing an auto tuning property to the vacuum conveying system and investigate the possible benefits. This thesis has investigated two main components for such an autotuner, the system design and optimisation strategy. The system design is critical for allowing process control and monitoring. In the optimisation the main challenge was to robustly produce good results for a wide variety of materials. Based on this a laboratory system was developed that monitored the capacity, efficiency, gentleness and robustness performance of the conveying. This information was then used in anevolutionary optimisation algorithm based on differential evolution to find the best settings for optimal performance. The results of this thesis indicate that such a system would offer benefits bothin terms of initial tuning and run-time performance. The suggested auto tuner is adaptable to a large set of materials with varying material properties and demands on the conveying process. The time required for the laboratory rig to optimise within a region with only small deviations in performance is comparable to that of a human operator. Also with the added benefit of never having to stop the optimisation process, thus constantly improving performance and counteracting environmental changes.