A Dynamic Monitoring System for Three-dimensional Operation of Blast Furnace Shape and Slag Skin Morphology based on PINNs -MPF

Abstract

With the deepening implementation of the "dual carbon" strategy, the steel industry is facing enormous pressure for green transformation.As the core equipment of ironmaking, the high temperature and high pressure "black box" characteristics inside the blast furnace seriously restrict refined operation and energy conservation and consumption reduction.The slag hanging state of the blast furnace cooling wall and the operating furnace type directly determine the service life and safety of the blast furnace.This paper proposes an intelligent dynamic monitoring system for blast furnace three-dimensional operation furnace shape and slag skin morphology, which integrates physical information neural networks (PINNs) and multiphase field (MPF) methods, to address the difficulties of traditional monitoring methods in real-time quantifying furnace wall thickness, slag skin detachment frequency, and effectively responding to the coupling effects of multiple physical fields.This study first uses a shared encoder and modality specific adapter to generate an adversarial network (SE-MSA-GAN) to solve the spatiotemporal alignment and denoising problems of multi-source heterogeneous data;Secondly, a three-dimensional heat transfer forward problem model based on PINNs MFF was constructed, and millisecond level high-precision reconstruction of the cooling wall temperature field was achieved through extended domain decomposition and pyramid training strategy;At the same time, a mathematical model based on water temperature difference was established to invert the thickness of slag skin and heat flux intensity in real time.The system integrates a high-precision sensor array and a digital twin visualization platform, which can dynamically display the three-dimensional status of the entire furnace and provide graded safety warnings.The experimental results show that the system can significantly improve the accuracy of fault diagnosis, providing strong technical support for the longevity maintenance and low-carbon smelting of blast furnaces.