Parallel Implementation Of Global Heat Source Factory Extraction Algorithms

While industry brings convenience to people's production and life,pollution also aggravates environmental problems.In addition,the development of industry often leads to disorderly expansion of industrial zones,unreasonable industrial layout and industrial structure.Therefore,it is of great significance to study the layout of industry.With the development of earth observation technology and the update iteration of satellite sensors and other equipment,the amount of data information captured by people is increasing.How to use satellite remote sensing data efficiently for factory research has become a research direction.Aiming at the problems of small research scope,long research cycle and low computational efficiency existing in the previous research on plant layout,this paper analyzed the background and significance,existing difficulties and future development trend of this research,studied the related technologies such as parallel computing,task partitioning and clustering analysis algorithm,and realized the global heat source plant in parallel environment.Extraction.Firstly,the K-means clustering analysis algorithm is optimized with the method of factory identification.Secondly,large-scale data-level task decomposition is implemented based on GeoSOT geographic partition coding.Based on the principle of dynamic allocation,a dynamic task tree for heat source plant extraction is constructed.Finally,in the parallel environment based on the message passing model MPI(Message Passing InterFace),the comparative experiments based on different computing resources,different data scales and different levels are carried out,and the computational efficiency and factory layout are compared and analyzed.This research uses literature review method and experimental method.The innovation lies in the extraction of heat source plants worldwide based on parallel computing.The experimental results show that the K-means clustering analysis algorithm is a feasible method to extract global heat source data from factories.Task decomposition based on GeoSOT geographic partition coding is suitable for large-scale parallel computing based on location relations.In addition,under the same conditions,the efficiency of parallel computing is much better than that of serial computing.By parallel computing,large-scale clustering computation can be processed quickly.