ML Core and Predictive Modeling
BKOAI’s ML Core streamlines the development of real-time predictive models through an accessible automated machine-learning workflow
Goals
- Simplify predictive-model development through automated machine-learning workflows.
- Make real-time predictions accessible through an operator-facing interface.
- Tailor predictive models to plant-specific operating conditions.
- Support operational decisions with accessible model outputs.
Challenges
- Fragmented industrial data requires preparation before it can support model development.
- Generic machine-learning models may not capture plant-specific operating behavior.
- Preparing data and configuring predictive models requires time and specialized expertise.
- Model outputs must be accessible within operators’ existing workflows to support their use.
Solutions
- Use template-driven contextualization to organize asset data consistently for analysis.
- Integrate predictive models with platforms such as Seeq and AVEVA PI to make outputs accessible.
- Tune machine-learning models using plant-specific characteristics and historical data.
- Provide an operator-facing interface for viewing predictions and model outputs.
Results
- Provided predictive alerts to support proactive operator responses.
- Made model outputs accessible through integrated dashboards.
- Standardized asset analytics through reusable templates.
.svg){kind=small}
.svg){kind=small}
.svg){kind=small}
