Granular Prediction and Dynamic Scheduling Based on Adaptive Dynamic Programming for the Blast Furnace Gas System

Area of Science:

• Industrial Engineering
• Process Control
• Artificial Intelligence

Background:

• Effective scheduling of byproduct gas systems is crucial for intelligent manufacturing and energy conservation in the steel industry.
• The blast furnace gas (BFG) system requires real-time dynamic scheduling to optimize operations.
• Existing methods may not adequately address the data fluctuations and system delays inherent in BFG operations.

Purpose of the Study:

• To propose a granular prediction and dynamic scheduling process for the blast furnace gas (BFG) system.
• To develop a method for real-time dynamic scheduling that enhances intelligent manufacturing and energy conservation.
• To accurately predict future trends in gas tank levels for optimal scheduling decisions.

Main Methods:

• Construction and description of information granules to reflect production data specificity.
• Establishment of a scheduling action network based on granular feature descriptions.
• Development of a tendency attenuation-based cost function considering system delays.
• Implementation of a reinforcement learning-based granulation and prediction process for gas tank levels.

Main Results:

• The proposed method demonstrates high accuracy in BFG system scheduling.
• Comparative experiments using industrial data validate the effectiveness of the approach.
• The method provides an effective solution for justified scheduling of the BFG system.

Conclusions:

• The adaptive dynamic programming approach with granular prediction offers a robust solution for BFG system scheduling.
• The developed method contributes to improved energy conservation and intelligent manufacturing in the steel sector.
• Accurate real-time scheduling of the BFG system is achievable with the proposed reinforcement learning and granulation techniques.