In the sintering process for rotary kiln, sintering temperature is the most important thermal parameter, which directly determines product quality and energy consumption. However, accurate prediction of sintering temperature is still infeasible. Traditional methods for temperature prediction often struggle to cope with the nonlinearity and dynamic variations inherent in the sintering process, leading to insufficient prediction accuracy. To address this issue, this study proposes a dynamic prediction method for sintering temperature in rotary kilns based on the DA-Seq2Seq (Dual-Stage Attention Sequence to Sequence) algorithm. The DA-Seq2Seq algorithm, with its unique dual-stage attention mechanism, effectively captures key features and long-term dependencies in time series data, thereby significantly enhancing the accuracy of sintering temperature prediction. To validate the effectiveness of the proposed method, extensive experiments were conducted on actual datasets of rotary kiln sintering temperature. The results demonstrate that the DA-Seq2Seq method significantly outperforms traditional prediction models in terms of prediction accuracy.