Spark-based imaging satellite task preprocessing parallelization method

摘要

Spark (RTM) based imaging satellite task preprocessing parallelization method includes: 101 performing orbit prediction parallelization design and decomposing a large-scale observation task into multiple small tasks, each small task being completed independently; and 102 implementing parallelized orbit prediction. Orbit prediction inputs of multiple satellites may be obtained, encapsulated into an RDD and output via an orbit prediction program. Elements in the RDD may be transmitted to the orbit prediction program by means of pipe. An output result from orbit prediction computing of each satellite in each computing time segment may be obtained and an orbit identification parameter may be used to distinguish between output results. Output results storage manner may be writing into a local file system, RDD or Redis (RTM) memory. Spark (RTM) cluster environment may be initialised and a cluster parameter may be configured. Preprocessing may be used in large‐scale satellite observation task planning or scheduling to convert a standardized requirement raised by a user, including e.g. target location, imaging quality or resolution and sun altitude angle, into meta‐tasks i.e. minimum imaging tasks that can be executed by a satellite.