Method Development for Device-Level Industrial Heat Source Identification Using Medium and High Resolution Satellite Images
SUN Shuang,JIANG Lei,LIU Baoxian,LU Haifeng,REN Huazhong,WANG Xinhui,LI Jinxiang
DOI: https://doi.org/10.11834/jrs.20232486
2024-01-01
Abstract:This study focuses on the internal thermal field distribution of a single large-scale factory to address the problems of insufficient refinement and large errors in the existing remote sensing monitoring of industrial heat sources.A device-level industrial heat source identification method based on medium and high-resolution satellite images is also proposed. The surface temperature is first obtained based on the dual-channel nonlinear split-window algorithm.Then,several 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.By using high-resolution satellite images,the range of the high-temperature device is determined,the difference in the accuracy of the recognition results of several methods is compared and analyzed,and the method with the best recognition effect is ascertained. (1)Each method has the ability to identify heat source devices.The boundary of the standard deviation grading of temperature,that is,1.5×,and the focal statistical analysis method is clearer.The accuracy rate and fall rate of the cell point are more than 85％.However,the omission rate of the general heat source device is also high at more than 50％.Thus,it is suitable for capturing the most important heat source device in the factory area.The identification results of the cold and hot spot analysis method are relatively stable,but the range is larger than that of the actual heat source device,and the non-falling rate is 23％.The result is more misidentified pixels,but more heat source devices can be captured.The identification results of the temperature 1 standard deviation grading method and the clustering and outlier analysis method are closely aligned.The accuracy rate is about 82％,and the omission rate is within 10％,which is more suitable for identifying heat source devices in the plant area.(2)Comprehensive analysis of the identification effect of the five methods shows that cluster and outlier analysis methods have good identification ability,the correct detection rate of identification results is more than 80％,the overall omission rate is less than 10％,and the identification results of each period can be obtained.This method is less affected by seasonal changes and is more suitable for identifying general heat source devices in the factory.(3)Judging by the characteristics of high-resolution satellite images,the heat source device mainly includes production device,circulating water,boiler device,storage warehouse,flare device,and tank farm.Among them,the production device accounts for about 55％,which is the main heat source device in the factory.The difference between the average temperature of each type of device and the factory area shows that the production equipment is also the main high-temperature area. Medium and high-resolution satellite remote sensing image data sources can effectively monitor the fine-scale heating units of industrial enterprises.They can provide technical support for environmental protection and management and industrial overcapacity reduction monitoring.