A self-learning state-aware rescheduling approach for robot-aided aero-engines fleet inspection. (October 2022)

• Record Type: Journal Article
• Authors: Song, Xinyi; Cheng, Ying; Ren, Jiawei; Tao, Fei
• Is Part Of: Journal of manufacturing systems. Volume 65(2022)
• Page Start: 802
• Page End: 810
• Publication Date: 2022-10
• Subjects: Rescheduling approach -- Robot-aided aero-engines fleet inspection -- Reinforcement learning -- Metaheuristic; Manufacturing processes -- Periodicals; Production engineering -- Data processing -- Periodicals; Robots, Industrial -- Periodicals; Production, Technique de la -- Informatique -- Périodiques; Robots industriels -- Périodiques; Electronic journals; 670.42
• DOI: 10.1016/j.jmsy.2022.11.010
• Languages: English
• ISSNs: 0278-6125

Abstract

Abstract: With the concept development of aircraft intelligent propulsion system, the use of intelligent robot swarms for aero-engines fleet inspection has become a trend. However, the uncertainty of aircraft arrival brings difficulties to the scheduling of inspection robots. Considering the effect of uncertainty on system state, a state-aware rescheduling approach is proposed for robot-aided aero-engines fleet inspection. In which, a system state evaluation considering both adequacy of inspection resources and urgency of inspection tasks is proposed. Reinforcement learning and metaheuristic algorithm are used to implement the self-learning rescheduling and scheduling plan solving respectively. Experimental results show the effectiveness of the proposed approach, which can overcome the short-sightedness of metaheuristic algorithm and achieve long-term improvement of the scheduling objectives. Highlights: For robot-aided aero-engines fleet inspection, a self-learning rescheduling approach is proposed considering system state. A system state evaluation method is proposed, including evaluation of the resource adequacy and the task urgency. Reinforcement learning is used for rescheduling self-learning and rescheduling plan is solved by metaheuristic algorithm. Experiments with different task arrival frequencies are designed to verify the effectiveness of the proposed approach.