On-board Autonomous Scheduling Based On Evolutionary Heuristics

As the hardware level continues to increase,the capabilities of imaging satellite are becoming stronger,and the development of satellite capabilities has led to higher application demands.Satellite has already had considerable computing ability,resource sensing ability,real-time task demand receiving ability and environmental detecting capabilities,which makes the on-board autonomous mission planning possible.In this context,it is possible for satellites to realize the transition from a simple instruction execution body to a combination of planning and execution,thus truly liberating the satellite's powerful detection capabilities and fulfilling more mission requirements.At present,the research on the autonomous mission planning method on the satellite is not complete,and the idea of re-planning is adopted.The ground is still the main body of mission planning,failing to reflect the independent status of the satellite.Therefore,under the general trend of satellite automation and intelligence,research on the star autonomous mission planning method in a completely autonomous environment can make up for the gap in this field,which has great theoretical and practical significance.As the latest generation of Earth observation satellites,agile satellites represent the future development direction of Earth observation satellites.Therefore,this paper selects agile Earth observation satellites as the research subject,and analyzes the characteristics of agile satellite capabilities and the implementation process of Earth observation.It is found that the traditional mission planning system has limitations such as long work cycle,poor effectiveness,and obvious conflicts between measurement and control resources when dealing with emergency tasks.The demand and necessity of on-board autonomous mission planning are analyzed and demonstrated in detail.Then,according to the knowledge method of scheduling theory,the autonomous satellite programming problem on agile satellites is standardized.From the perspective of engineering practice,some reasonable assumptions are put forward.Considering the timing dependence characteristics of agile satellite mission planning,the corresponding constraint satisfaction model is built according to the multi-stage triggered autonomy task planning mode.Considering that the autonomous satellite programming problem on the agile satellite is an NP-Hard problem in the field of telemetry,heuristic algorithms are usually used to solve the problem in the actual engineering field,but the general heuristic rule algorithm often cannot obtain satisfied results.Based on the characteristics of the problem,this paper proposes an evolutionary method based on genetic programming,which can automatically evolve suitable constructive heuristic rules for any given autonomous satellite on-board autonomous mission planning problem instance.Individuals in genetic programming are mathematical functions corresponding to heuristic rules,and then by using a time-based constructive heuristic algorithm to map heuristic rules into an observation scheme of a specific scene,and obtain corresponding rule adaptation values,and then inspire The rule evolves to find the best rule for the scene,and then through the generalization of the scene set,the rule set for autonomous task planning on the star can be obtained.Finally,the simulation experiment is designed with reference to the actual satellite parameters.The evaluation indexes commonly used in the field of satellite Earth observation are selected.The multi-group analogy experiments are used to compare and analyze the on-board autonomous mission planning method and other methods based on evolution rules.The results prove the effectiveness of the on-board autonomous mission planning method based on evolution rules designed in this paper in different task scale scenarios.