SmartDIL is a Promine tool that uses Artificial Intelligence to predict and reduce ore losses and dilution tonnage.
What is SmartDIL ?
SmartDIL is an artificial intelligence tool for predicting and controlling dilution and ore losses in underground mines
An easy to use solution to predict and reduce dilution and losses. SmartDIL takes into account the historical data of the mining operation. It is also able to simulate and compare various methods in order to reduce dilution. The client can then measure the impacts of the changes in order to improve his performance according to the methods he has chosen.
How it works ?
SmartDIL collects and normalizes historical stope data (dimensions, rock type, drill and blast data, dilution results, etc.). From this data, an artificial intelligence model is trained and able to predict ore dilution and loss for given planned stopes. The more data is made available, the more precise and accurate the modelled impact will be.
The principle components of SmartDIL:
Estimation of ore dilution and loss at a given stope
A simple and practical web interface
Detailed reports of the stope
Calculation of the fluctuations in ore dilution and loss with time
Automatic sharing of stope data to SmartDIL
Communication with CAD tools (Promine)
Better control of dilution and losses = increased profitability
Facilitate the work of the engineer while improving the results.
Keep and improve expertise on dilution management within the organization.
Reduction of the time before finding an optimal solution.
Reduction of the environmental impact.
The process of dilution reduction relies on analyzing stope data with artificial intelligence to optimize the results. Consequently, the results are appended to a database to increase the precision of the next process.
of data with artificial intelligence
of dilution of a planned stope
of the design in order to reduce dilution
of the stope and application of design
of the results of the stope to the database
The main parameter influencing the accuracy of the prediction:
· The number of mined sites given to AI
The other parameters having an influence (not necessarily in this order):
· Quality of the extracted data.
· Site with similar properties.
· Adjustment of the weight of certain parameters by the users.
· Varieties of extracted data: blasting data, geological data, field support, etc.
The accuracy of the prediction will be greater with a greater number of mined sites. The accuracy of the prediction can be represented with the following curves :