Multi-Model- and Soft-Transition-Based Height Soft Sensor for an Air Cushion Furnace

Area of Science:

• Industrial Engineering
• Process Control
• Sensor Technology

Background:

• Accurate measurement of strip floating height in air cushion furnaces is crucial for production quality and efficiency.
• Harsh operating conditions (high temperature, high pressure) prevent direct measurement.
• Multiple strip floating states complicate accurate modeling.

Purpose of the Study:

• To develop a multi-model soft sensor for estimating strip floating height.
• To address challenges posed by different floating states and measurement limitations.
• To enhance process prediction and control in air cushion furnaces.

Main Methods:

• State identification using adaptive k-nearest neighbors and principal component analysis.
• Development of a hybrid model incorporating mechanistic, double-random forest, and data gravitation-based soft-transition models.
• Validation using a self-developed air cushion furnace experimental platform.

Main Results:

• Successfully divided floating states and developed state-specific predictive models.
• The multi-model soft sensor accurately estimated strip floating height across different states.
• Demonstrated effectiveness and value for process prediction and control.

Conclusions:

• A novel multi-model soft sensor approach effectively estimates strip floating height in air cushion furnaces.
• The method integrates state identification and soft transitions for robust prediction.
• This research offers significant value for optimizing air cushion furnace operations.