What is it all about? The fact is that Norilsk Nickel, together with the Russian Hackers hackathon community, will hold an online hackathon from April 16 to 18, dedicated to finding the best solutions for optimizing processes in the non-ferrous metals mining industry. Developers, data scientists, analysts and managers and representatives of other specialties are invited to participate in the hackathon. Teams from 3 to 5 people can take part in the hackathon. In order to better understand the essence of the hackathon, we will tell you about the company's technologies, and then - the details of the event itself, with a link to registration.
Briefly (actually not) about ore dressing at the dressing plant in Norilsk Nickel
More precisely, it will focus on processing plants and ways to optimize production using high technologies.
As for the enrichment itself, its fundamental principles have not changed for decades. Technology - yes, beneficiation efficiency - yes. Principles - either do not change at all, or insignificantly.
Imagine you have mined ore. The metal content (no matter what, iron or nickel) is not very high there. In order to obtain this metal, the starting material needs to be enriched - namely, to increase the concentration of the metal. The increase in metal content is due to the fact that what is not metal is separated from the ore and goes to the tailing dump (in fact, production waste).
This is achieved in different ways, as an example, we will cite the Talnakh enrichment plant:
- Stage 1. Ore is crushed using mills. After that, a simple process of separating the conditioned product from oversized elements is carried out. There is a grate at the exit from the mill, everything that does not pass the grate is sent to the mill again.
- Stage 2. Flotation - This method uses the separation of waste rock and useful product. The flotation method is based on the difference in the wettability of the initial components with water.
- Stage 3. Thickening, after which the product of enrichment, together with water, goes to metallurgical plants.
The presented process is rather simplified, everything is more complicated, but it is suitable for understanding the essence of metal enrichment. So, each of the three stages can be optimized in order to increase the amount of useful product at the output. And this is where new technologies come into play.
Big data in enrichment
New technologies are implemented by the development team for all three stages, including related processes.
Now, 14 different initiatives have already been implemented, of which 2 have been implemented, this is production management in automatic mode, 7 is at the testing stage and 5 is the R&D stage.
Duration of projects - R&D 2-3 months, pilot 2-3 months, implementation takes 2-6 months.
The company began to form a team of specialists in 2019, then it consisted of only 2 people. Now it is already 5 people - 1 PM and 4 DS.
Technologies used:
- Classic machine learning
- Computer vision
Current stack:
- Python, SQL
- 2 servers in the data center of Norilsk
- 2 servers in production (1 for computer
- view)
Here are some visual examples of concentrator projects.
As for the plans, this is mainly the scaling up of already proven technologies - now they are used in two sections of the process of the plant out of 10. In total, the technologies are tested at three processing plants.
Not only enrichment
Inefficiency and the search for their fixes can be everywhere - from technology to management processes, the search for such problems and their solution is the company. This leads to increased production efficiency and improved financial performance.
So, in addition to enrichment, the company is gradually introducing machine learning and big data in areas such as occupational safety and health. These are tracking the health level of employees, predictive models of the dynamics of seasonal diseases, and several other projects.
Another major project is Data Lake. It is a technological platform for solving human resource (HR) tasks
The Data Lake provides a unique opportunity to create an extended employee profile and to carry out management and forecasting at a new level in one of the strategic areas for business - in personnel management (HR).
In this case, information can be collected from various internal systems. The implementation of data-driven approaches in HR analytics is one of the most promising trends in AI transformation. The main goal here is to improve the efficiency of employees, improve industrial safety, prevent accidents, etc.
Ok, it's clear with technology, but what about the hackathon?
It consists of two main tracks:
Foam Party - determination of the flow rate of foam with metal and other flotation parameters from the video to optimize the operator's work in production.
Forecasting continuous employment - creating a model for predicting sick leave of employees in one of the production workshops of Norilsk Nickel using anonymized data about employees and their environment.
The total prize fund of the hackathon is 500,000 rubles: 250,000 rubles for each track. Plus, each participant gets a merch. Moreover, for the People's Choice Award, participants receive AirPods Pro, and Yandex.Plus, Okko, Gmoji promo codes are also available.
What the hackathon gives to the participants:
- data driven .
- - .
- .
" We live in an interesting time - artificial intelligence penetrates into all spheres, even into such a seemingly conservative as heavy industry. As an example, at one of Norilsk Nickel's factories, part of the flotation process is already controlled by a digital twin, while actively enriching data computer vision technologies are used, including for determining the size of crushed ore.
Work in this direction is very interesting and in its own way unusual - it includes both deep immersion in machine learning algorithms and in the technology of the production process. a wide range of people and the hackathon is a great opportunity to learn how industrial ML works using real examples", - commented Anton Abrarov, Head of data science projects at Norilsk Nickel. The
technological partner of the hackathon is SberCloud, which will provide participants with everything they need to train models: ML Space - a platform for full cycle ML development and collaboration of DS teams, as well as the cloud infrastructure of the hackathon itself. powerful Russian supercomputer “Christofari.” You can
register for the hackathon from March 17 to April 14.
Well, in order to answer the main questions about the hackathon, in addition to the information above, we asked a few questions to one of the organizers - Alexander Malyshev from Russian Hackers .
Who, how and why did you decide to take and organize Hackathon with "Norilsk Nickel"?
. , , . , «» Hack.Moscow, , - .
?
, – . , – . , , , , -, «». , Russian Hackers.
- / , ?
– , , Angular. data-science – .
, ? ? ?
Data Science. , . , .. 50% . , , , , . , «», Russian Hackers SberCloud.
? « »?
As organizers, we first of all set the task of telling what tasks are in a fairly closed industry from the point of view of Data Science. It will be cool if the teams get a job at Norilsk Nickel or launch a pilot with them, but here, as in any B2B, you need to have strong willpower to bring a prototype from a hackathon to a real result in production. Let's see how it goes.
Well, we are waiting for the participants of the hackathon. Just in case, we once again indicate the link for registration .