Abstract: In the process of developing and utilizing coal resources, a large amount of data is generated, and this data contains a lot of potentially valuable information. Making full use of the massive coal quality data accumulated in the process of coal development and utilization and mining the hidden information can generate new information and apply it to social production and construction. Areas with advantageous coal resources under different geological conditions will present a clustering phenomenon of data distribution. Four parameters of raw coal, including moisture (M_ad), ash yield (A_d), volatile matter (V_daf) and total sulfur (S_{t, d}) of the Taiyuan Formation in the six major coal fields in Shanxi Province are selected. the raw data is preprocessed using SOM+K- Means algorithm processing, and the read data is first processed based on the self-organizing neural network SOM, and the result obtained is used as the second stage k-means clustering analysis algorithm for further processing. According to the relevant national standards, the two types of data are displayed on the map according to the different quality of the raw coal, and the advantageous areas are delineated. The results of data mining show that the dominant areas of high-quality coal and medium quality coal account for 90.1% and 24.1% in the first and second clusters, respectively, indicating that the first cluster has higher quality coal than the second cluster. This proves the possibility of data mining algorithm coal quality big data analysis, expands the use of coal quality data, and further provides new ideas for the use and development of coal quality databases.