Edge-Computing and Machine-Learning-Based Framework for Software Sensor Development.

Area of Science:

• Chemical Engineering
• Computer Science
• Artificial Intelligence

Background:

• Machine learning (ML) models are increasingly used for quality assurance in materials science.
• Existing ML models often lack solutions for continuous operation, maintenance, and version control.
• Difficult-to-measure laboratory variables in the chemical industry present a challenge for traditional methods.

Purpose of the Study:

• To present a framework for the development, operation, and lifecycle management of ML-based software sensors.
• To leverage ML and edge computing for innovative solutions in the chemical industry.
• To enable continuous product quality forecasting and stable plant operation.

Main Methods:

• Utilizes ML algorithms and cloud services within an enterprise environment.
• Integrates Industrial 4.0 technologies: Internet of Things (IoT), edge computing, cloud computing, ML, and AI.
• Develops an information system structure and methodology based on quality-assurance laboratory data.

Main Results:

• The developed framework supports the full lifecycle of ML models, including development, version control, validation, and maintenance.
• It addresses challenges in continuous ML model development and parameter tuning.
• The framework enables continuous performance monitoring and data enhancement for ML models.

Conclusions:

• The proposed framework ensures that quality-assurance engineers have access to the most accurate, data-driven, and up-to-date ML models.
• It offers innovative solutions for managing software sensors and difficult-to-measure variables in the chemical industry.
• The research contributes to effective quality control, stable production, and efficient laboratory work through advanced ML integration.