Citizen data scientists working in asset-intensive intensive industries today are faced with managing multiple types of models in their organization’s operations. These frequently include custom Python models for failure or anomaly detection, machine learning models from libraries like Tensor Flow and Scikit-Learn, First Principles engineering models (e.g., AspenTech’s HYSYS® and Aspen Plus®), hybrid models that combine first principle physics-based models and domain knowledge, Excel-based models, models that consume other services (e.g., weather API), and any other models on-prem or in the cloud that expose a callable endpoint. Without a singular infrastructure to execute, manage and share results from these disparate models, the citizen data scientist is relegated to administrative oversight rather than strategic decision-making.
To enable citizen data scientists and other advanced users to focus their time on the highest value activities, AspenTech, as part of its latest V14 software release, has equipped Mtell®, its award-winning asset reliability and performance technology, with the capability to integrate external models. This “Bring Your Own Model” approach lets citizen data scientists:
- Leverage the data science team to collaborate with maintenance and process engineers
- Utilize Mtell’s architecture including historian connectivity, data collection, agent scheduling, security and logging
- Use the Mtell Alert Management and Mtell notifications systems to share all model results
- Manage all live data processing and the connection to the historian.
“To effectively manage large asset infrastructures and successfully support business and sustainability objectives, it’s essential that data scientists have a comprehensive, real-time, sharable view into operations across the entire enterprise,” said Don Busiek, VP and General Manager of AspenTech’s APM business. “Using the newest release of Mtell, this means easily integrating all model sources, including those from AspenTech’s competitors, to provide a singular, efficient infrastructure.”
AspenTech’s “Bring Your Own Model” capability is now available with the latest aspenONE® software release, V14.