Online concentrate band position detection for a spiral concentrator using a Raspberry Pi

- Organization:
- International Mineral Processing Congress
- Pages:
- 10
- File Size:
- 792 KB
- Publication Date:
- Jan 1, 2014
Abstract
"Spiral concentrators offer a cost effective way to separate valuable minerals from gangue by displacing lighter materials to the outside, and heavier materials to the inside of the spiral. The separation can be seen as visible concentrate bands and splitters are set at the bottom of the spiral to effectively split the ore into various densities. As there are typically many spirals in a concentrator plant, it becomes ineffective to continually adjust the splitter positions manually. Vermaak et al. (2008) proposed to use image recognition to detect the concentrate bands, thereafter, creating a model predicting the positions under certain input conditions. This model could then be used for optimisation and/or control of splitter position or band position. In contrast an online concentrate band detection system would provide real time concentrate band positions without losing accuracy due to model generalisation. These systems are usually expensive and time consuming to develop, due to a lot of centralised features required (image capturing, image processing and/or feedback to a controller). This study took advantage of the recently developed Raspberry Pi minicomputer to perform rapid prototyping of an online image processing sensor. The Raspberry Pi, and accompanying camera module, offers a portable Linux environment which can natively run the required image processing algorithms. Furthermore, extensive use of OpenCV was made to reduce development time, by relying on existing and proven algorithms to detect the concentrate bands. The Raspberry Pi also incorporates generic input/output ports which can be used in future work to provide the signal to a controller. In conclusion, the Raspberry Pi was successfully used to perform real time concentrate band position detection on a spiral concentrator. This solution is more general than an experimentally identified input-output model and may be used for control or optimisation studies in a wide array of spiral concentrators."
Citation
APA:
(2014) Online concentrate band position detection for a spiral concentrator using a Raspberry PiMLA: Online concentrate band position detection for a spiral concentrator using a Raspberry Pi. International Mineral Processing Congress, 2014.