Intelligent methods in mining environment
In this project, machine vision systems are studied and applied to mineral enrichment process. Practical implementations are located in the Pyhäsalmi mine, some 500 km's north from Helsinki.
Machine vision provides new information that can be used for example in: characterization of flotation froth, crushed ore particle size distribution measurements and colour/spectral measurements. With this new information, it is possible to further study the process itself which is quite complex and difficult to model. However, with proper models and sophisticated controllers, the process can be controlled better and thus economical savings can be achieved. Flotation Cell
The main objectives of the project are
- better control over flotation circuits
- improved colour measurements
- improved particle size distribution measurement
Better control over flotation process can be achieved by introducing new measurements, improving measurement accuracy and by better controllers. One example are the laser-based froth distance measurements. They provide new information that is used to calculate froth thickness and to compensate the error that is introduced to other image-based measurements due to fluctuation of the distance between the froth surface and the camera. Another way to improve measurement accuracy is to combine the machine vision information coming from different flotation cells. This information is measured simultaneusly and thus gives a more coherent picture of the whole process. Improved controllers will, among other things, utilize this new information.
Colour measurements seem to be very important in mineral flotation. The current way of measuring colour is by means of a RGB-colour camera. However, standard cameras are designed to measure colour in the same wavelength range as human eye. Introduction of new spectrophotometers enables a more thorough investigation of the froth and slurry colour with respect to colour depth and to wavelength range.
Particle size distribution is measured from a conveyor belt by using a properly placed light source and a monochrome-camera. The measurement point is located in the mine, some 1400 meters below ground level. The important thing in this research area is to be able to predict what the size distribution will be when the ore reaches the surface. This takes usually 2-3 days. Recent studies have indicated that the particle size distribution present in the mine is changed toward fines during transportation. This change will be modelled and taken into account in the prediction algorithms.
The research is financed by Tekes and by the industrial partners.
- Keywords: mining, mineral flotation, image analysis, data mining, soft sensors, process control, process monitoring
- Duration: 2005-2006
- Research area: Process Control, Operations & Maintenance
- Related projects: ChaCo, VÄSY, MINMO
- Publications (total: 11)
- Heikki Koivo (person in charge)
- Jani Kaartinen (project manager)
- Heikki Hyötyniemi
- Olli Haavisto
- Timo Roine
- Antti Tolonen
- Martti Larinkari: Particle size distribution of crushed ore - measurement and management (2004)
- Olli Ojala: Enchancement of X-ray analyzer measurements utilizing Deterministic-Stochastic Subspace Identification (2004)
- Jani Kaartinen: Data Acquisition and Analysis System for Mineral Flotation (2001)