Genetic programming improves adaptive scheduling rules for changing factories
Lifelong Learning of Scheduling Rules with Surrogate-assisted Genetic Programming for Dynamic Flexible Job Shop Scheduling
Artificial Intelligence
Summary
Scheduling tasks in factories with many machines is tricky because job types and arrivals keep changing. The authors address how to keep machine scheduling rules smart and flexible over time without forgetting what was learned before. They use a genetic programming approach that adapts rules for new tasks while remembering past knowledge. Their method also estimates how similar new tasks are to old ones to reuse useful rules better. Tests show their approach handles changes well and beats other methods in flexible scheduling.
What this means in practice
- •For manufacturing planners: Create scheduling rules that adapt over time to varying job arrivals and shop conditions without losing previous effectiveness.
- •For factory automation engineers: Build factory software that updates scheduling strategies dynamically in response to new production tasks using genetic programming.$Commercial implications: Enables development of adaptive factory scheduling software that can be sold to manufacturing automation vendors for improved efficiency.
Authors
Luyao Zhu, Fangfang Zhang, Yi Mei, Ke Chen, Mengjie Zhang
Abstract
Genetic programming (GP) has been widely used to learn scheduling rules for dynamic flexible job shop scheduling (DFJSS). However, trained rules tend to only generalise to tasks of similar scales and distributions, and when task characteristics change, learned rules often suffer significant performance degradation. Moreover, fine-tuning pre-trained rules on new tasks often leads to catastrophic forgetting, where prior knowledge is lost. This motivates the need for GP methods that can adapt to new tasks while preserving past knowledge, leading to a lifelong learning paradigm. However, lifelong learning in GP for DFJSS remains largely under-explored. To address this gap, this paper proposes a surrogate-assisted GP method for lifelong learning to handle sequential heterogeneous DFJSS tasks with job or batch arrivals. The framework includes three key components: an adaptive two-stage training strategy to balance plasticity and stability, a task-similarity-based weighting mechanism to regulate knowledge reuse, and an improved surrogate model with distance constraints to enhance estimation accuracy. Experimental results show that the proposed method effectively mitigates catastrophic forgetting while maintaining strong adaptability, outperforming several compared methods and demonstrating the potential of lifelong GP in dynamic scheduling environments.