SOM neural network fault diagnosis method of polymerization kettle equipment optimized by improved PSO algorithm

Area of Science:

• Chemical Engineering
• Artificial Intelligence
• Process Control

Background:

• Real-time fault diagnosis and optimization monitoring are critical for the polyvinyl chloride (PVC) resin production process, specifically for polymerization kettles.
• Existing methods may lack the efficiency and accuracy required for complex industrial processes.

Purpose of the Study:

• To propose an effective fault diagnosis strategy for polymerization kettles in PVC production.
• To enhance the real-time monitoring and fault identification capabilities of the process.

Main Methods:

• A fault diagnosis strategy based on the self-organizing map (SOM) neural network is developed.
• The particle swarm optimization (PSO) algorithm, with a novel dynamical adjustment method for inertial weights, is used to optimize SOM structural parameters.
• A nonlinear mapping from symptom sets to fault sets is established for fault pattern classification.

Main Results:

• The proposed PSO-SOM fault diagnosis strategy effectively establishes a mapping between polymerization process data and fault patterns.
• Simulation experiments using industrial on-site historical data demonstrate the strategy's effectiveness in fault diagnosis.
• The optimized SOM network accurately classifies fault patterns in polymerization kettle equipment.

Conclusions:

• The developed PSO-SOM fault diagnosis strategy meets the real-time fault diagnosis and optimization monitoring requirements for PVC production polymerization kettles.
• This approach offers a robust and efficient solution for identifying and monitoring faults in industrial chemical processes.
• The integration of PSO and SOM provides a powerful tool for enhancing process safety and efficiency.