Investigation of the Generalisation Ability of Genetic Programming-evolved Scheduling Rules in Dynamic Flexible Job Shop Scheduling

TL;DR

This paper addresses the generalisation of GP-evolved routing and sequencing rules for Dynamic Flexible Job Shop Scheduling (DFJSS) when tested on instance types that differ in scale, parameters, and distributions. It employs a simulation-based DFJSS framework and a multi-tree GP setup to evolve scheduling rules, evaluating them across systematic variations in problem scale, utilisation, due date factor, batch size, and data distributions, with 30 repetitions per setting. The key finding is that generalisation is strongest when training and test instances share similar decision-point distributions or when training uses larger scales, while large disparities in scale or distribution severely degrade performance; the number and distribution of decision points emerge as central factors. The study highlights the need for GP rules with stronger, more generalisable patterns and suggests future work combining GP with lifelong learning to cope with heterogeneous DFJSS environments, enabling robust scheduling in dynamic real-world settings.

Abstract

Dynamic Flexible Job Shop Scheduling (DFJSS) is a complex combinatorial optimisation problem that requires simultaneous machine assignment and operation sequencing decisions in dynamic production environments. Genetic Programming (GP) has been widely applied to automatically evolve scheduling rules for DFJSS. However, existing studies typically train and test GP-evolved rules on DFJSS instances of the same type, which differ only by random seeds rather than by structural characteristics, leaving their cross-type generalisation ability largely unexplored. To address this gap, this paper systematically investigates the generalisation ability of GP-evolved scheduling rules under diverse DFJSS conditions. A series of experiments are conducted across multiple dimensions, including problem scale (i.e., the number of machines and jobs), key job shop parameters (e.g., utilisation level), and data distributions, to analyse how these factors influence GP performance on unseen instance types. The results show that good generalisation occurs when the training instances contain more jobs than the test instances while keeping the number of machines fixed, and when both training and test instances have similar scales or job shop parameters. Further analysis reveals that the number and distribution of decision points in DFJSS instances play a crucial role in explaining these performance differences. Similar decision point distributions lead to better generalisation, whereas significant discrepancies result in a marked degradation of performance. Overall, this study provides new insights into the generalisation ability of GP in DFJSS and highlights the necessity of evolving more generalisable GP rules capable of handling heterogeneous DFJSS instances effectively.

Figures (3)

• Figure 1: The whole process of GP for DFJSS and an example of a GP individual.
• Figure 2: Workload distributions and overlap ratios among instances with different number of jobs.
• Figure 3: Workload distributions and overlap ratios among instances with different utilisation levels.