Online estimation of slag volume inside grate kiln pellet plant by stereo vision and machine learning

Purpose and goal:

The project has partially succeeded in developing an improved and quantitative method for determining the volume of the slag build-up. The project has made advances in optics and image quality for cameras in demanding environments and developed image processing methods that create information valuable for the calculation of slag volume. Development would need to continue to achieve a more reliable and realistic calculation of the volume, with particular focus on identifying common points in the right/left images used for mapping points to a 3d space.

Expected results and effects:

The project has taken the system for measuring slag volume to a higher technical level, a step towards a fully developed system. A 3D measurement in the demanding environment of a pellet plant is difficult, but is still considered possible and also value-creating should the development be continued. Should development continue and the system reaches a sufficiently high technical level, this would be a unique and very high-tech system for monitoring hot processes where slag formations can occur.

Approach and implementation:

The project group met once a week and carried out the practical work between meetings. Project meetings helped participants to solve problems, generate ideas, and direct the week´s work. The steering group met once a month to monitor progress and oversee the project.

During parts of the project, most of the practical work was assigned to one or two people. This was a necessary arrangement, but led to lost time in the project because we had handovers to new employees a total of four times in the project (an unusually high turnover of key personnel in a project of this kind).