Adaptive Digital Twin Modeling with Control: Integration of Extended Kalman Filter-Based Recursive Sparse Nonlinear...

Area of Science:

• Industrial Process Control
• Digital Twin Technology
• System Dynamics

Background:

• Digital twins offer significant potential for industrial process simulation, monitoring, and control.
• Current implementations face challenges like long development times, reduced model accuracy, and limited interactivity.

Purpose of the Study:

• To propose a comprehensive digital twin development framework addressing key implementation challenges.
• To enhance the effectiveness, accuracy, and interactivity of digital twins in industrial processes.

Main Methods:

• Developed a framework integrating digital twin identification, real-time model updating, and advanced process control.
• Utilized sparse identification of nonlinear dynamics for offline model identification, reducing development time.
• Employed the extended Kalman filter for real-time model accuracy mitigation.
• Integrated updated models into model predictive control for optimized control inputs.

Main Results:

• Demonstrated reduced digital twin development time while maintaining model fidelity.
• Successfully mitigated diminishing model accuracy through real-time updates.
• Enhanced control input optimization and digital twin interactivity.
• Validated the framework's advantages through an industrial case study and simulation examples.

Conclusions:

• The proposed digital twin framework effectively addresses critical challenges in industrial applications.
• The integration of sparse identification, extended Kalman filter, and model predictive control offers a robust solution.
• The approach enhances the practical utility and performance of digital twins in industrial settings.