To address the limitation of existing industrial heat sources extraction methods based on high-temperature data are difficult to identify low temperature (<500 K) and small-scale factories, this paper proposes an industrial heat sources identification method that integrates thermal anomaly features from long-term remote sensing image. Firstly, for the surface temperature data inverted from satellite remote sensing image data, a global threshold combined with temperature-feature dynamic detection approach is used to extract potential industrial thermal anomaly points. Secondly,based on the annual scale, we integrated results from multiple periods of industrial thermal anomaly points to construct an industrial heat sources identification model combining DBSCAN clustering and logistic regression.Finally, taking Tangshan City, China as the research area, industrial heat sources identification was conducted to verify the accuracy of the model. The results showed that the accuracy of the industrial heat sources identification model was 86%, which significantly improved the ability to identify low-temperature and small-scale factories compared to the extraction results based on VIIRS Active Fire products.