Unscented Kalman Filter-Based Robust State and Parameter Estimation for Free Radical Polymerization of Styrene with...

Area of Science:

• Chemical Engineering
• Polymer Science
• Process Control

Background:

• Free radical polymerization of styrene (FRPS) involves complex systems with uncertain parameters.
• Parameter uncertainty and changes due to altered conditions or faults complicate process monitoring and control.
• Traditional unscented Kalman filter-based state and parameter estimation (UKF-SPE) struggles with timely detection and tracking of parameter variations.

Purpose of the Study:

• To develop a robust state and parameter estimation (UKF-RSPE) method for FRPS that addresses changing model parameters.
• To enhance the accuracy and timeliness of parameter and state estimation in dynamic polymerization systems.
• To improve process monitoring and control for FRPS under varying conditions.

Main Methods:

• Proposed a UKF-based robust SPE (UKF-RSPE) method incorporating a parameter testing criterion using hypothesis testing and moving windows.
• Implemented a gradient descent method with an adaptive learning rate for iterative parameter updates upon detection of changes.
• Applied the UKF-RSPE method to a continuous stirred tank reactor for FRPS.

Main Results:

• The proposed UKF-RSPE method successfully detected and tracked parameter changes in the FRPS process.
• The method demonstrated faster parameter tracking compared to traditional UKF-SPE.
• Achieved more accurate state estimation due to improved parameter identification.

Conclusions:

• The UKF-RSPE method is effective and robust for state and parameter estimation in FRPS with variable parameters.
• The approach enhances process monitoring and control by providing timely and accurate parameter tracking.
• Validated through experimental application in a jacketed continuous stirred tank reactor.