Spatial Decomposition-Based Fault Detection Framework for Parabolic-Distributed Parameter Processes.

Area of Science:

• Process control and automation
• Dynamic systems analysis
• Fault detection and diagnosis

Background:

• Distributed parameter processes (e.g., heat, fluid) require robust fault detection for safe operation.
• Existing data-driven methods fail to capture process dynamics, limiting their use in transient conditions.
• Current model-based methods lack redundancy by relying solely on boundary measurements.

Purpose of the Study:

• To develop a robust, model-based fault detection approach for distributed parameter processes.
• To address limitations of existing methods by incorporating both boundary and in-domain measurements.
• To enable real-time fault detection without extensive data collection or offline training.

Main Methods:

• A real-time fault detection filter (FDF) was designed based on a model-based perspective.
• Theoretical analysis was conducted for parameter selection and threshold computation.
• A time-varying threshold was implemented to prevent false alarms during transient phases.

Main Results:

• The proposed FDF effectively detects faults in distributed parameter processes.
• The method eliminates the need for large datasets and offline training.
• Successful application in a hot strip mill cooling system validated its industrial potential.

Conclusions:

• The developed fault detection filter offers a robust and efficient solution for dynamic industrial processes.
• The approach enhances operational safety and efficiency by providing reliable fault detection.
• The FDF demonstrates significant potential for real-world industrial applications, particularly in complex systems like cooling processes.