The Prediction of the Gas Utilization Ratio based on TS Fuzzy Neural Network and Particle Swarm Optimization

Area of Science:

• Industrial Engineering
• Artificial Intelligence
• Data Science

Background:

• Gas Utilization Ratio (GUR) is a critical metric for blast furnace (BF) energy efficiency.
• Existing GUR prediction methods lack accuracy, hindering effective energy management.

Purpose of the Study:

• To develop a novel, accurate data-driven model for predicting BF GUR.
• To enhance online blast furnace distribution control through improved GUR prediction.

Main Methods:

• Utilized a Takagi-Sugeno fuzzy neural network (TS-FNN) for GUR prediction.
• Employed Particle Swarm Optimization (PSO) to optimize TS-FNN parameters, reducing initial parameter errors.
• Applied the Box-plot method for robust data preprocessing, handling non-normally distributed industrial data.

Main Results:

• The PSO-optimized TS-FNN model demonstrated superior prediction accuracy compared to standalone TS-FNN and Support Vector Machine (SVM) models.
• The proposed model effectively predicts GUR in complex industrial environments.

Conclusions:

• The novel data-driven approach integrating PSO and TS-FNN offers a significant advancement in GUR prediction accuracy.
• This method provides an effective solution for real-time blast furnace energy consumption control.