Text

IDEA: Identifying key variables for monitoring of production processes in automotive industry: A deep learning method

The project IDEA aims to automatically identify key variables for process monitoring by means of deep learning.

Concluded

Start

2021-04-01

Conclusion

2021-12-31

Main financing

Co-financing

Collaboration partners

Project manager at MDU

No partial template found

In automotive industry, the scale and complexity of production processes are increasing, which leads to high safety and production quality demands. Process monitoring is crucial to ensure proper process operations and fast detection of anomaly and hazardous events. As lots of variables of the process are being collected and stored, it gives rise to the problem of data with too high dimension, the so called “curse of dimensionality”.

The project IDEA aims to automatically identify key variables for process monitoring by means of deep learning. The point of departure is that not all the underlying variables are relevant in monitoring for detection of anomaly or hazard. Excluding irrelevant variables from the detection model will not only reduce the complexity but also substantially improve model accuracy. IDEA will be carried out with close collaboration between Mälardalen University and Volvo Truck.

Purpose of the project

The project IDEA aims to automatically identify key variables for process monitoring by means of deep learning

Project objectives

Develop efficient learning algorithms to identify the subset of key variables as well as data clusters that reflect normal process behaviours.


This research relates to the following sustainable development goals