Development of Intelligent Process Surveillance
Reference number | |
Coordinator | ACOSENSE AB |
Funding from Vinnova | SEK 500 000 |
Project duration | November 2013 - June 2014 |
Status | Completed |
Important results from the project
Acosense develops and sells Acospector Acoustic Chemometer, a Clamp-on instrument measuring properties of complex fluids in the process industry. In this project, methods were developed to automatically identify deviating observations with the purpose of gradually incorporating normal data into the model and thereby improving the model to get increased prediction quality. An algorithm based on the clustering-method DBSCAN was implemented and was verified to be able to detect deviating observations. In addition, support for process status visualization has been implemented.
Expected long term effects
The project was completed with good results. Apart from most of the initial goals, some additional results were achieved. Among these results are improved measurement precision, more effective use of data, and increased understanding of the interactions between measurements, instrument hardware and measurement object. For Acosense AB this has resulted in further increased product quality, more effective deployment processes and increased robustness. Finally, the project also led to identifying and formulating a new product for which development has begun.
Approach and implementation
The project was conducted as a collaboration between Acosense AB and Fraunhofer Chalmers Centre for Industrial Mathematics. The work was carried out in close cooperation, with regular visits to the offices of the two partners. A master thesis at KTH was also integrated into the project.