Abstract: A thorough analysis of the characteristics of the industrial structure is the key to deeply understanding the region-al economic dynamics and promoting the transformation and upgrading of industries. The in-put-output table (IOT) provides detailed inter-industry flow data for the analysis of industrial structure characteristics and is used to quantify the correlation between input-output among industrial sectors. However, due to the inherent spatial characteristics, dynamic change characteristics and complex correla-tions of data, it poses a severe challenge to the analytical methods for revealing industrial structure pat-terns. For this purpose, this paper designs a spatiotemporal multi-dimensional visual analysis system for industrial structure characteristics based on IOT. Firstly, based on the IOT data of all provinces across the country, an industrial input-output network among the industrial sectors of each province is constructed. Then, the triangular structure with clear economic implications was utilized to quantify the functional roles of industries. A multi-dimensional feature vector was constructed and combined with the t-SNE algorithm for dimensionality reduction projection, thereby visually presenting the distribution of industrial parts with similar industrial structure patterns. Secondly, design multiple interactive and convenient associated views to display the characteristics of industrial structure models in different industrial sectors, thereby helping users quickly perceive the laws and differences in the evolution of industrial structure models over time. Thirdly, design and integrate a visualization system to achieve multidimensional visual analysis of indus-trial structure features in time and space, supporting users to interactively explore the characteristics of industrial structure models and their temporal evolution laws from multiple perspectives. Finally, using real IOT data from 31 provincial-level administrative regions in my country, a case study was conducted on the cross-regional differences in coal mining and washing, as well as the industrial structure evolution of metal mining and tourism in Zhejiang Province, and two experts in the field evaluated the method from the per-spectives of the effectiveness of the triangular structure feature extraction method, the practicality of mul-ti-view collaborative visualization design, and the convenience of human-computer interaction design. The evaluation results verified the effectiveness of the proposed method and system in solving the problems of industrial structure feature mining and spatiotemporal evolution analysis.