Transforming data mapping in mining

Diagram showcasing the data mapping workflow in MASTERMINE

By LIBRA AI Technologies

Data mapping is a critical process for managing and transforming data, ensuring seamless integration across different sources. In the context of the MASTERMINE project, data mapping supports the gathering and utilisation of mining-related data for time-series classification. This allows mines to standardise and automate the integration of sensor data, weather patterns, metal prices, and environmental indicators into the MASTERMINE platform.

By leveraging new and existing mining sensor data, the goal is to create a demonstrative model that can automatically classify time-series data from a new mine and map it to the appropriate fields within MASTERMINE.

What is Data Mapping?

Data mapping is the process of matching data fields from different sources to a common structure, ensuring consistency, accuracy, and usability. In MASTERMINE, this process is vital for:

• Structuring raw sensor-generated mining data.

• Identifying time-series classifications for various operational KPIs.

• Facilitating seamless integration of new mines into the platform.

With AI-powered automation, data mapping becomes faster, more scalable, and adaptable to various mining operations.

Why Apply Data Mapping?

Implementing data mapping in MASTERMINE brings several advantages:

• Efficient Data Integration: Standardises diverse data sources, including environmental, operational, and market data.

• Automation Through AI: Reduces manual work by leveraging machine learning to suggest mappings.

• Improved Decision-Making: Ensures high-quality, structured data for analytics, monitoring, and forecasting.

• Scalability: Enables new mines to integrate their data into MASTERMINE quickly and efficiently.

Data Mapping in MASTERMINE

To support time-series classification, the MASTERMINE data mapping pipeline consists of the following key steps:

1. Data Acquisition

• Mining sensor data is collected from various sources (IoT devices, environmental sensors, market indicators).

• Relevant KPIs and schemas are defined to align with MASTERMINE’s platform requirements.

       2. Model Development and Training

• Time-series classification models are trained using diverse datasets.

• The methodology explores ML and deep learning (DL) techniques to extract features, embeddings, or alternative data transformations.

       3. Model Testing

• The model is validated with new sensor data to ensure it generalises across different mining operations.

• Performance metrics such as accuracy, precision, and recall are evaluated.

      4. Model Deployment and Usage

• Mines submit their data files to a secure storage system in a tabular format.

• The system analyses the data, suggests potential mappings, and prepares it for review.

• Users confirm or update the suggested mappings and integrate the data into the MASTERMINE database for further use.

Demo: Sensor Data Processing in a Mine

To illustrate the real-world impact of data mapping, a demonstration was conducted using sensor data from an operational mine.

Demo Overview

• Mine-Specific Data: Temperature, CO₂ levels, and humidity from underground tunnels.

• Process Steps:

1. Uploading Data: Raw sensor readings were uploaded to the Data Bucket.

2. Data Mapping Execution: The system automatically suggested mappings for each sensor reading.

3. User Validation: Engineers reviewed the AI-generated mappings and finalized the configurations.

4. Integration into MASTERMINE – The structured data was ready to be added to the platform for real-time monitoring and analysis.

Key Results

• 80% correctly mapping sensor readings to predefined fields.

• 50% reduction in time required for data integration in a new mine.

The data mapping process enhances efficienty and scalability in mines digitalization exploting the potencial of AI in minining industry

What’s Next?

To further refine the data mapping process, MASTERMINE will:

• Expand training datasets to improve model accuracy across different mining conditions.

• Enhance automation to minimize manual intervention in the mapping process.

• Optimise UI/UX for better user experience in data selection and validation.

By continuously iterating and refining the approach, MASTERMINE aims to build a fully automated, intelligent data mapping system that accelerates mine’s digitalisation.