A method for establishing visual images of models of battery status transforms data of one or more characteristic parameters collected in a period of time into multiple images composed by normalized pixels corresponding to the data. The multiple images are superimposed with one another to form variant image(s) with variant areas and further moving trends of the variant areas. The visualization of data and superimposition of the visualized images facilitates correlation and variation analysis among characteristic parameters of the battery module. The present invention transform data analysis to image analysis, and therefore, the present invention facilitates following analysis and model training, which can adapt to environmental changes, and indicate battery status and battery events efficiently and accurately.