Knowledge-Driven Memetic Algorithm with Forgetting Mechanism for the Energy-Aware Flexible Job-Shop Scheduling Problem with Variable Machining Speeds

Against the backdrop of global energy conservation, emission reduction, and sustainable development, energy-aware scheduling in manufacturing workshops has become a growing focus. However, existing research on the energy-aware flexible job-shop scheduling problem (EFJSP) with variable machining speeds primarily considers energy consumption at the job level, neglecting the energy heterogeneity of individual operations under different constraints, a factor of key environmental significance. To address this problem, this paper innovatively incorporates processing quality constraints to define upper limits on variable speeds. The study develops a mathematical model for the speed-adjustable flexible job shop scheduling problem considering operation-level energy consumption. Considering limitations in current multi-neighborhood structure selection, we propose a knowledge-driven memetic algorithm with a forgetting mechanism (KDMA-FM). The model's effectiveness is validated through comparisons with the exact solver OR-Tools on a simplified system. Furthermore, KDMA-FM's performance is evaluated against existing algorithms across instances of varying scales. Results demonstrate that KDMA-FM outperforms benchmark algorithms in over 80% of instances and achieves notable energy savings.