Abstract: Quality prediction modeling based on deep learning has become one of the mainstream methods in the field of soft sensor, it can more effectively address complex problems such as nonlinearity and high-dimensionality compared to traditional statistical analysis and shallow machine learning. Transformer, a deep learning model based on the self-attention mechanism, offers a new research avenue for industrial quality prediction due to its unique advantages in sequence modeling. This paper systematically reviews the technological evolution and applications of Transformer in the industrial process quality prediction. Initially, the limitations of classic deep learning models in dynamic industrial processes are discussed. By analyzing the ability of the Transformer model's self-attention mechanism to model dynamic global features of industrial processes, its fundamental application principles in industrial process soft sensor problems are highlighted. Subsequently, the main improvement directions of the classic Transformer method in existing literature, including Crossformer, iTransformer, Informer, PatchTST, and Autoformer, are summarized, along with their application cases in complex real-world industrial processes. Finally, current challenges in industrial quality prediction are discussed, and future research directions are prospected.