首页 > , Vol. , Issue () : -
In view of the problems of insufficient refinement and large errors in the existing remote sensing monitoring of industrial heat sources, this paper focuses on the internal thermal field distribution of a single large-scale factory, and proposes a device-level industrial heat source identification method based on medium and high-resolution satellite images. . That is, the surface temperature is first obtained based on the dual-channel nonlinear split-window algorithm, and then a variety of spatial statistical analysis methods are used to identify candidate high-temperature areas, and the location of high-temperature devices in the plant is determined by multi-temporal superposition analysis of the identification results, and then combined with high-resolution satellite images. Determine the range of the high temperature device, compare and analyze the difference in the accuracy of the recognition results of several methods, and determine the method with the best recognition effect. The conclusions are as follows: (1) Multi-temporal-hot spot analysis and multi-temporal-clustering and outlier analysis methods have better recognition results, and the identified pixels fall in rates above 80%, and the overall omission rate and the omission of production equipment rate is less than 10%. (2) A total of 139 high-temperature devices were identified in the test area, including 56 production devices, followed by a total of 11 storage warehouses, and less than 10 of other types of devices. The production device is the main heating device. It is simple and effective to identify industrial heat sources at the device level based on multi-phase-cold-hot spot analysis and multi-phase-clustering and outlier analysis methods, and the results are accurate and reliable. It can monitor changes in industrial production activities at the device level. Provide technical support services.