Coal Preparation » Fine Coal
Researchers have developed an on-line, vision-based prototype instrument to measure coal flotation froth properties.
Traditionally, optimal metallurgical performance in coal flotation circuits has been difficult to achieve due to the wide variety of process disturbances.
Although on-line analysis instruments have been available for the past 20 years, they are generally expensive to purchase and maintain, do not adequately detect process behaviour within flotation cells, and many contain nucleonic sources which present potential health and safety issues.
Over the past decade, the cost of image analysis hardware and software has become significantly lower, and new graphic file formats make it possible to handle and display images at resolutions beyond the capability of the human eye. Potentially, application of computer-based vision technology offers a number of advantages compared with other on-line analysis techniques:
- The measurement is non-intrusive and non-contact (no interference with the process)
- No mechanical parts are involved, such as sampling and de-aeration systems, so there are minimal maintenance requirements
- Low capital cost, and
- High sampling frequency
The project's objectives are to:
- Develop image analysis algorithms which can characterise flotation froth in terms of froth structure, bubble size and horizontal froth velocity (pulling rate)
- Specify, develop and test a prototype machine vision system which can be used as a froth analyser to improve manual and automatic flotation control.
Researchers investigated the use of digital image analysis techniques to make quantitative measurements of surface froth characteristics, and applied the measurements to the evaluation and control of a coal flotation cell. A texture spectrum algorithm was developed to measure bubble size and froth structure, while an object-matching algorithm called pixel tracing was developed to measure froth velocity. The prototype instrument was successfully developed and evaluated on a Microcel flotation column at Peak Downs coal preparation plant.
Conclusions that can be drawn from the use of texture spectrum techniques are:
The texture spectrum method offers some important advantages compared with the other image analysis techniques:
- The regional background noise of an image is reduced to a minimum
- The conceptual approach is simple, reducing the amount of calculation required
The texture spectrum of a froth image provides a unique fingerprint of the froth. As a result, a classification algorithm using patterns of texture spectra could be developed to recognise froth structures. This algorithm was based on a set of standard froth structures which were collected under different process conditions at Peak Downs mine.
The height of the middle peak of the texture spectrum can be employed to measure the average bubble size of the froth image.
Conclusions that can be drawn from the development of the pixel tracing technique are:
Pixel tracing is suitable for automatic analysis where computational cycles are fixed because the technique is independent of the moving objects present in the scene. In traditional image analysis techniques where each object in the scene is traced individually, the computation time required for each analysis is different (in the worse case, it would take an infinite time).
The technique is simple but robust. It can measure froth speed and be used to predict the direction of the flow. However, in practice, only the principal direction of the froth has been considered in the analysis.
Where To From Here
The future work for this research project may be concentrated in:
- Developing an advanced prototype instrument
- Incorporating the prototype instrument into a flotation control strategy
- Applying the techniques developed to other types of coal flotation cell, and
- Using the digital image analysis to study the sub-processes occurring within the froth