ADAV - Automatic validation of data from industrial processes

Important results from the project

The feasibility study has been designed to analyze the potential of using emerging multivariate technologies to automatically validate quality and integrity in industrial data before it is used as a basis for decision making. The study shows potential for reduced costs in steel industry and Bio Process in the order of ~ 500 MSEK / year. The result of the study forms the basis for an application (Diary Number: 2018-02212) for a continuation through a small FUI project under the SIP PiiA Calling Spring 2018.

Expected long term effects

The project has investigated, quantified and described the value of the idea through concrete calculation examples. Different detection techniques have been investigated and an autoassociative neural network is estimated to be of particular interest and potential. Firstly, because the method´s requirements for training data fit well with an industrial data set and partly because the method can provide some insight into which process data are the basis for the detected deviation. The project partners assess the results for feasibility and value creation of the idea as high.

Approach and implementation

Erasteel Kloster AB and GE Healthcare Bio-Sciences AB have analyzed the potential of the idea for value creation in their respective operations. Sprymer AB has acted as coordinator in the project and evaluated different detection technologies in relation to the idea and feasibility of applying them in industrial environments. For increased cooperation, the parties have met for discussions at project meetings that took place both at parties and via the internet.