ChaCo

The characterization of flotation froth structure and colour by machine vision

Keywords: machine vision, colour image analysis, flotation froth, process modelling, process control

Flotation is a common industrial method by which valuable minerals are separated from waste rock. Flotation cells are instrumented with various sensors but still certain variables e.g. size and form of the bubbles and the colour of the froth are only visually observed by the operator. The aim of the project is to characterise and measure these variables using machine vision techniques and to use this information in process control.

The objectives of the project are: 1. To analyse the mineral concentration from the colour image of the froth 2. To design an on-line froth analyser 3. To develop process models and control methods 4. To test the results at industrial flotation plants.

During the first project year HUT has collected image and spectrum data and analysed it with statistical and neural network methods. A permanent teleoperateable machine vision system has been designed and installed at a plant. The system enables remote data collection, software updates and additions and produces measurements of the froth characteristics to the process computer and thus available to operators.

During the second project year more complex image analysis algorithms have been installed, tested and tuned for on-line operation at the industrial test set-up. The froth classification studies led to the identification of three froth classes: stiff, wet and dry. First a method based on predetermined typical feature vectors for each class was constructed and implemented. Next a neural network method was found to lead up to congruent results.

During the third project year the methods were developed so that they are easily chosen and utilised in the system. A summary study of the classification methods was carried out. The dependences of the image and process data were studied and a control based on that was developed. The performance of the control was studied and it was found highly beneficial for the plant

Publications

• Jani Kaartinen and Heikki Koivo,
  "Machine vision based measurement and control of zinc flotation circuit,"
  Studies in Informatics and Control, Vol. 11, no. 1, pp. 97-105, 2002.
• H Hyötyniemi, R Ylinen and J Miettunen,
  "Soft computing at a flotation plant,"
  in Industrial Applications of Soft Computing - Paper, mineral and metal processing industries, K Leiviskä (ed.), Heidelberg, Germany: Physica-Verlag, 2001, pp. 133-143.
• J Kaartinen and V Hasu,
  "Konenäön soveltaminen mineraalien vaahdotuksen mittaukseen ja säätöön,"
  Automaatioväylä, no. 1, s. 10-12, 2001.
• Jani Kaartinen,
  Data acquisition and analysis system for mineral flotation,
  Master's thesis, Department of Automation and Systems Technology, Helsinki University of Technology, Espoo, Finland, 2001. (Helsinki University of Technology, Control Engineering Laboratory, Report 126).
  Electronic publication
• G Bonifazi, V Giancontieri, A Meloni, S Serranti, F Volpe, R Zuco, H Koivo, J Hätönen, H Hyötyniemi, A Niemi, P Sipari, H Kuopanportti, R Ylinen, I Heikkilä, S Lähteenmäki, J Miettunen, O Stephansson, W Wang and Carlsson,
  "Characterization of the flotation froth structure and color by machine vision (ChaCo),"
  in Proceedings of the XXI International Mineral Processing Congress, P Massacci (ed.), The Netherlands: Elsevier Science B.V, 2000, Vol. C/C8a pp. 39-49.
• V Hasu, J Hätönen and H Hyötyniemi,
  "Analysis of flotation froth appearance by design of experiments,"
  in Future Trends in Automation in Mineral and Metal Processing, S.-L Jämsä-Jounela and E. Vapaavuori (eds.), Finland: IFAC, Finnish Society of Automation, 2000, pp. 470-474.
• H. Hasu Hyötyniemi, J V. Hätönen and R Ylinen,
  ""Data mining" for mining data,"
  in Proceedings of the XXI International Mineral Processing Congress, P Massacci (ed.), The Netherlands: Elsevier Science B.V, 2000, Vol. A/A3 pp. 39-45.
• H Hyötyniemi and V Hasu,
  "Independent components of expertise,"
  in STeP 2000 - Millennium of Artificial Intelligence, Vol. 3: 'AI of Tomorrow', H Hyötyniemi (ed.), Helsinki, Finland: Finnish Artificial Intelligence Society, 2000, Vol 3, pp. 37-44.
• H Hyötyniemi and R Ylinen,
  "Modeling of visual flotation froth data,"
  Control Engineering Practice, Vol. 8, no. 3, pp. 313-318, 2000.
• H Hyötyniemi, R Ylinen and J Miettunen,
  "AI in practice: Case study on a flotation plant,"
  in STeP 2000 - Millennium of Artificial Intelligence, Vol. 2: 'AI of Today', H Hyötyniemi (ed.), Helsinki, Finland: Finnish Artificial Intelligence Society, 2000, Vol 2, pp. 159-166.
• Vesa Hasu,
  Design of Experiments in Analysis of Flotation Froth Appearance,
  Espoo: Helsinki University of Technology, Control Engineering Laboratory, 1999. (Helsinki University of Technology, Control Engineering Laboratory, Report 114).
• H Hyötyniemi,
  Software: GGHA Toolbox for Matlab,
  1999.
• J Hätönen,
  Image analysis in mineral flotation,
  Espoo: Helsinki University of Technology, Control Engineering Laboratory, 1999. (Helsinki University of Technology, Control Engineering Laboratory, Report 116).
  Electronic publication
• J Hätönen, H Hyötyniemi, G Bonifazi, S Serranti, F Volpe and L.-E Carlsson,
  "Using PCA in controller strategy design for a flotation process,"
  in Proceedings of the 14th IFAC World Congress 1999, H.-F Chen, D.-Z Cheng and J.-F. Zhang (eds.), Beijing, P.R. China: IFAC, 1999, Vol. N, pp. 385-390.
  Electronic publication
• J Hätönen, H Hyötyniemi, J Miettunen and L.-E Carlsson,
  "Using image information and partial least squares method to estimate mineral concentrations in mineral flotation,"
  in Proceedings of the Second International Conference on Intelligent Processing and Manufacturing of Materials - IPMM'99, J. A Meech, M. M Veiga, M. H. Smith and S. R LeClair (eds.), Honolulu, Hawaii: IEEE, 1999, Vol. 1, pp. 459-464.
  Electronic publication
• A. J Niemi, H Hyötyniemi and R Ylinen,
  "Image analysis and vision systems for processing plants,"
  in Proceedings of the Second International Conference on Intelligent Processing and Manufacturing of Materials - IPMM'99, J. A Meech, M. M Veiga, M. H. & LeClair Smith and S. R. (eds.), Honolulu, Hawaii: IEEE, 1999, Vol. 1, pp. 11-20.
  Electronic publication
• H Hyötyniemi,
  "Structure from data: AI approaches to systems modeling,"
  in Human and Artificial Information Processing, P Koikkalainen and S Puuronen (eds.), Helsinki, Finland: Finnish Artificial Intelligence Society (FAIS), 1998, pp. 31-40.
• H Hyötyniemi and R Ylinen,
  "Modeling of visual froth data,"
  in Preprints of the IFAC Symposium on Automation in Mining, Mineral and Metal Processing 1998, J Heidepriem (ed.), New York: Elsevier Science, 1998, pp. 309-314.
• H Koivo, R Ylinen and P Sipari,
  "Machine vision research in Control Engineering Laboratory of the Helsinki University of Technology,"
  in Konenäköseminaari - Machine Vision Seminar 26.-27.5.1998, Innopoli, Espoo, Finland, Helsinki, Finland: Suomen Automaatioseura ry, Vision Club of Finland, 1998, pp. 61-61 (abstract).
• R Ylinen, H Hyötyniemi, E Ikonen and U Kortela,
  "A new Hebbian type learning algorithm applied to sensor fusion,"
  in COSY Annual Joint Workshop 1998, Ohrid, Macedonia: ESF, 1998, p. 9.