Research progress on perception, prediction and control technologies for burn-through point in iron ore sintering

The burn-through point （BTP） is a key thermal parameter reflecting the arrival of combustion zone at the material layer bottom and the burning through status of layer. Its stability directly affects sinter quality, fuel consumption, and sintering production stability. Due to factors such as raw material fluctuation, changes in bed permeability, distribution uniformity, delayed characterization of thermal state, and multivariable coupling, the BTP is difficult to accurately determine and stably control using a single signal. For BTP online identification, advance prediction, and optimal control, related research has gradually developed a trajectory from multi-source sensing including wind box temperature fields, flue gas components, machine tail images, and infrared thermography, to predictive modeling using shallow neural networks, ensemble learning, and deep learning, and further to single objective stable BTP control, coordinated control of BTP and bin material level, and intelligent optimal control under complex operating conditions. This paper systematically reviews the research progress in BTP sensing and identification methods, prediction models, and control strategies, with a focus on analyzing the applicable scenarios and engineering limitations of different technical approaches. Existing studies show that BTP control is evolving from empirical judgment and local feedback regulation towards multi-source sensing, feedforward prediction, and collaborative optimization. However, further breakthroughs are still needed in multi-source heterogeneous data fusion, cross condition generalization, model interpretability, and long term online application. The analysis results presented in this paper can support the development of intelligent BTP control systems and the optimization of the sintering process.