1080 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
pricing are tied to the ore body knowledge—this is the big-
gest value driver for an AI-driven application.
Examples of AI Applications
Figure 11 and Figure 12 provide examples of AI appli-
cations in mining. Simpler AI systems are suitable for
repetitive tasks in mining systems, such as machine vision,
automation, data review, and equipment fault-finding.
The connected AI systems are more solutions-focused and
are complex, such as optimizing the mine plan schedule,
maximizing productivity, and minimizing energy and water
footprint.
Increasing the data resolution of the ore during the
exploration phase has the biggest opportunity. Subsequent
Table 4. Steps to accelerating AI application in the mining industry
Steps Key Focus Tasks
1 Data Availability • Undertake a complete review of all data sources to build integrated mining operations (geology,
mine, process, asset, constraints)
• Build a high-level static model of the entire process to understand the extent and sources of various
data streams—the model can be in the form of a simulation or digital twin of the process
• If data is lacking, then install instrumentation or measurement to capture the pertinent information
(more drill hole test work data, the instrumentation on the process, asset diagnostics instruments,
etc…)
2 Build Models /
Relationships
• Build operational relationships using the available data
• Leverage on first principle models and empirical relationships to build an understanding of the
mining process
• Benchmark, the model, outputs with actual measurements
• Identify gaps in the model outcomes—these areas will define the opportunity to build an AI
platform
3 Consolidate all
unit processes
and drive
constraints
analysis.
• Define the constraints for the model /AI platforms to aim at—i.e., energy and water consumptions,
operating costs, ESG metrics, etc.…
• Prioritize the constraints—as a start, begin with the process production (overtime, the AI platform
should learn to reprioritize the mine depending on the changing constraints (dynamic constraints)
• Review, consolidate and improve the entire mining value-change—this means more instruments,
more refines data sources and improved AI models
Figure 11. Building simpler AI systems (Chandramohan, 2023)
pricing are tied to the ore body knowledge—this is the big-
gest value driver for an AI-driven application.
Examples of AI Applications
Figure 11 and Figure 12 provide examples of AI appli-
cations in mining. Simpler AI systems are suitable for
repetitive tasks in mining systems, such as machine vision,
automation, data review, and equipment fault-finding.
The connected AI systems are more solutions-focused and
are complex, such as optimizing the mine plan schedule,
maximizing productivity, and minimizing energy and water
footprint.
Increasing the data resolution of the ore during the
exploration phase has the biggest opportunity. Subsequent
Table 4. Steps to accelerating AI application in the mining industry
Steps Key Focus Tasks
1 Data Availability • Undertake a complete review of all data sources to build integrated mining operations (geology,
mine, process, asset, constraints)
• Build a high-level static model of the entire process to understand the extent and sources of various
data streams—the model can be in the form of a simulation or digital twin of the process
• If data is lacking, then install instrumentation or measurement to capture the pertinent information
(more drill hole test work data, the instrumentation on the process, asset diagnostics instruments,
etc…)
2 Build Models /
Relationships
• Build operational relationships using the available data
• Leverage on first principle models and empirical relationships to build an understanding of the
mining process
• Benchmark, the model, outputs with actual measurements
• Identify gaps in the model outcomes—these areas will define the opportunity to build an AI
platform
3 Consolidate all
unit processes
and drive
constraints
analysis.
• Define the constraints for the model /AI platforms to aim at—i.e., energy and water consumptions,
operating costs, ESG metrics, etc.…
• Prioritize the constraints—as a start, begin with the process production (overtime, the AI platform
should learn to reprioritize the mine depending on the changing constraints (dynamic constraints)
• Review, consolidate and improve the entire mining value-change—this means more instruments,
more refines data sources and improved AI models
Figure 11. Building simpler AI systems (Chandramohan, 2023)
