Abstract: During the sintering ignition process, the airbox pressure needs to be maintained within the target range to keep the furnace chamber at a slightly negative pressure. In actual sintering production, the valve opening required for the blast box pressure within the target range is difficult to predict, and incorrect adjustment of the blast box valve often makes the internal pressure of the blast box not meet the sintering production requirements and makes it difficult to achieve the sintering micro-negative pressure ignition. To address this problem, a sintering bellows valve prediction model is proposed using genetic algorithm (GA) to optimize the BP neural network, which is trained with actual sintering data from a steel mill, and the prediction results of several prediction models are compared. The experimental results show that the prediction effect of the GA-BP prediction model is better than that of the traditional BP prediction model and the fitted prediction model, and the model can predict the valve opening required for the blast box pressure to reach the target range more accurately, which provides reliable theoretical support for the intelligent control of the blast box valve opening at the sintering site and can well meet the production requirements.