Abstract: Studying the spatio-temporal distribution law and evolution characteristics of stress in coal mining stopes is the key to establishing disaster early warning indicators and implementing effective prevention and control. Building upon the recent research progress of our research team in digital modeling and intelligent mining of coal mines, this research focuses on constructing a stope stress database and achieving the visual output of its evolution characteristics. To be specific, a digital reconstruction method for stope stress through data acquisition, fidelity preservation, transmission, and normalization processing was proposed based on the multi-source data of stope stress, and a multi-source data fusion and efficient storage database equipped with functions such as intelligent database retrieval, information output, and cloud platform query was constructed. Aiming at the problems such as data discreteness in the stope stress database, a data interpolation method based on micro-element piecewise Fourier transform was introduced, and a three-dimensional interpolation algorithm of "generating lines from points, generating surfaces from lines, and generating volumes from surfaces" was put forward, achieving the continuity representation of multi-source discrete data. Meanwhile, three types of stope stress inversion models, namely linear, polynomial, and exponential, were constructed based on key points and key regions, and a three-dimensional visualization output inversion algorithm for the stope stress evolution characteristics was developed. Furthermore, physical experimental research on the stope stress evolution characteristics was conducted, through which the construction of a multi-source database of stope stress and the visualization output of its evolution characteristics were achieved at the laboratory scale. Taking the typical structure of Da'anshan Coal Mine of Beijing Haohua Energy Group and the 3105 mining face of Mataihao Coal Mine in Ordos, Inner Mongolia as the engineering geological background, the fine interpolation of the in-situ stress field and the visualization inversion of mining-induced stress based on micro-seismicity positioning and hydraulic support stress were carried out, achieving the dynamic visualization output of stope stress at the field scale. The research results show that the collection and fidelity, effective fusion, and classified storage of stope data are realized through multi-source data missing filling, repetition elimination, data noise reduction, and format conversion, and the database can conduct precise and rapid query and location of stope data based on retrieval conditions. The stope stress interpolation algorithm based on Fourier transform improves accuracy by 88%, 82%, and 74% respectively in one-dimensional, two-dimensional, and three-dimensional stress continuity represen-tations. By acquiring and analyzing the support pressure and micro-seismicity data of the mining face, the dynamic evolution process of the mining-induced stress of the mining face was reproduced at the field scale. The distribution of multi-source monitoring data is in good agreement with the inversion results, verifying the accuracy of the inversion results.