Shipboard aircraft sortierecovery online scheduling method based on deep reinforcement learning

Abstract

The invention discloses a shipboard aircraft sortierecovery online scheduling method based on deep reinforcement learning, relates to the field of shipboard aircraft departure recovery scheduling, andsolves the problem of large-scale shipboard aircraft departure recovery online scheduling on an aircraft carrier deck. According to the method, a shipboard aircraft launch recovery scheduling model is constructed by abstracting a launch recovery process into a Markov decision process, a current state of a shipboard aircraft group is taken as input and taking a scheduling behavior as output and establishing a feature vector with a weight as a reward function. In order to obtain a safe and efficient scheduling strategy, a multi-target scheduling strategy is established, and a scheduling model is trained by using a deep Q learning network by taking shipboard aircraft shipboard displacement, task scheduling time, shipboard conflict times and equipment utilization rate as targets. Experimentaltest results show that the algorithm can quickly process emergencies and does not affect subsequent task execution, and meanwhile, a scheduling strategy with high safety and flexibility is obtained.

Description

technical field [0001] The present invention relates to the field of recovery and scheduling of carrier-based aircraft, and mainly relates to a scheduling decision-making method optimized by applying a deep reinforcement learning algorithm to solve the online scheduling problem of large-scale carrier-based aircraft when considering various constraints and emergencies. Background technique [0002] An aircraft carrier embodies the capability of a country to conduct maritime operations, and the combat capability of an aircraft carrier mainly depends on the dispatch capability of the carrier-based aircraft on the aircraft carrier. For other reasons, the dispatch and recovery process of carrier-based aircraft is carried out in a space-limited and dangerous environment, so it needs to rely on limited resources such as catapults, landing runways, and operators to work at high speed. How to rationally utilize the limited deck space and support resources in harsh operating environme...

AI Technical Summary

Problems solved by technology

Due to the large amount of calculation of the heuristic algorithm, most scholars at home and abroad only study the scheduling problem of small batches of carrier-based aircraft groups, which is unreasonable in the actual combat process

At the same time, in the process of flight deck operation planning and scheduling, for example, when carrier-based aircraft cannot take off on time, catapults or traction equipment cannot work, and support tasks cannot be completed in time, etc. Scheduling strategy, will lead to very serious consequences

Therefore, the decision-making method must have the ability to solve real-time conditions flexibly and quickly, while the traditional heuristic algorithm not only spends a lot of time in scheduling calculations, but also lacks real-time performance and has poor online scheduling capabilities.

Images