在當代製造與服務場域中，排程問題廣泛存在於產線管理、訂單處理、資源配置等應用場景，其複雜性與多樣性導致其被歸類為 NP-hard 問題。面對不同特性與需求的排程任務，單一排程演算法往往難以全盤適用，因此選擇適切的解法策略成為提升排程效能之關鍵。過去研究已提出多種啟發式與元啟發式演算法，如遺傳演算法（GA）、禁忌搜尋（Tabu Search）、螞蟻演算法（ACO）、模擬退火法（SA）、粒子群最佳演算法（PSO）、束搜尋法（Beam Search）等，皆於特定問題中展現良好效能，惟其表現高度依賴於排程實例本身的特徵。為提升排程系統於多樣實例下的決策能力，本研究提出一套整合神經網路（ Neural Network, NN）與指針網路（Pointer Network）之架構。研究中透過神經網路學習排程實例的結構化特徵與啟發式演算法迭代資訊並以指針網路作為策略，實現各式排程問題近似解，本實驗將以 NSGA-II 作為標竿學習對象，且以零工式生產排程問題(Job-Shop Scheduling Problem, JSSP) 排程案例進行模擬驗證，評估本架構於不同問題規模與特性下之整體表現與適應能力。

Scheduling problems are ubiquitous across manufacturing and service industries, spanning production planning, order fulfillment, and resource allocation. Due to their complexity and diverse constraints, such problems are typically classified as NP-hard, making it difficult for a single algorithm to perform optimally across all scenarios. While a variety of heuristic and metaheuristic algorithms—such as Genetic Algorithm (GA), Tabu Search, Ant Colony Optimization (ACO), Simulated Annealing (SA), Particle Swarm Optimization (PSO), and Beam Search—have shown promising results in specific contexts, their performance is often highly sensitive to the characteristics of individual scheduling instances. To enhance the adaptability of scheduling systems, this research proposes an integrated framework that leverages Neural Networks (NNs) and Pointer Network. The proposed approach leverages NNs to learn both the structural features of scheduling instances and the iterative information generated by heuristic algorithms, while employing Pointer Networks as the policy to produce approximate solutions for various scheduling problems. Experimental validation is carried out using NSGA-II as the benchmark for learning, and by simulating the Job-Shop Scheduling Problem (JSSP) under a flexible manufacturing scenario. The evaluation examines the framework’s overall performance and adaptability across different problem scales constraints.

Bahdanau, D., Cho, K., & Bengio, Y. (2015). Neural machine translation by jointly
learning to align and translate. arXiv preprint arXiv:1409.0473.
Dauzère-Pérès, S., & Paulli, J. (1997). An integrated approach for modeling and solving the
general multiprocessor job-shop scheduling problem using tabu search. Annals of Operations research, 70(0), 281–306.
Deb, K., Pratap, A., Agarwal, S., & Meyarivan, T. (2002). A fast and elitist multiobjective
genetic algorithm: NSGA-II. IEEE Transactions on Evolutionary Computation, 6(2), 182–197.
Dorigo, M., & Gambardella, L. M. (1997). Ant colony system: A cooperative learning
approach to the traveling salesman problem. IEEE Transactions on Evolutionary Computation, 1(1), 53–66.
Echeverria, I., Murua, M., & Santana, R. (2025). Self-evaluation for job-shop scheduling.
arXiv preprint arXiv:2502.08684.
Glorot, X., Bordes, A., & Bengio, Y. (2011). Deep sparse rectifier neural networks. In
Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics (AISTATS 2011), 315–323.
Glover, F. (1986). Future paths for integer programming and links to artificial intelligence.
Computers & Operations Research, 13(5), 533–549.
Goldberg, D. E. (1989). Genetic algorithms in search, optimization, and machine learning.
Addison-Wesley, 1989(102), 36.
Hamilton, W. L., Ying, R., & Leskovec, J. (2017). Inductive representation learning on
large graphs. Advances in neural information processing systems, 30.
Han, Y., Huang, G., Song, S., Yang, L., Wang, H., & Wang, Y. (2022). Dynamic neural
networks: A survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(11), 7436–7456.
Kirkpatrick, S., Gelatt, C. D., Jr., & Vecchi, M. P. (1983). Optimization by simulated
annealing. Science, 220(4598), 671–680.
Leray, P., & Gallinari, P. (1999). Feature selection with neural networks. Behaviormetrika,
26(1), 145-166.
Li, X., & Gao, L. (2016). An effective hybrid genetic algorithm and tabu search for flexible
job shop scheduling problem. International Journal of Production Economics, 174, 93–110.
Loshchilov, I., & Hutter, F. (2017). Decoupled weight decay regularization. arXiv preprint
arXiv:1711.05101.
Nazari, M., Oroojlooy, A., Snyder, L., & Takác, M. (2018). Reinforcement learning for
solving the vehicle routing problem. Advances in Neural Information Processing Systems, 31, 9839–9849.
Pinedo, M. L. (2016). Scheduling: Theory, algorithms, and systems (5th ed.). Springer.
Schuster, M., & Paliwal, K. K. (1997). Bidirectional recurrent neural networks. IEEE
Transactions on Signal Processing, 45(11), 2673–2681.
Tsai, C.C. (2024). A bi-objective genetic algorithms for solving a job shop scheduling
problem with material constraint and parallel batching when minimizing makespan and total number of tardy stage-outs. Master’s thesis. National Central University..
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, Ł., &
Polosukhin, I. (2017). Attention is all you need. In Advances in Neural Information Processing Systems, 30, 5998–6008.
Veličković, P., Cucurull, G., Casanova, A., Romero, A., Liò, P., & Bengio, Y. (2018).
Graph Attention Networks. arXiv preprint arXiv:1710.10903.
Vinyals, O., Fortunato, M., & Jaitly, N. (2015). Pointer Networks. Advances in neural
information processing systems, 28, 2692–2700.
Wang, T., Payberah, A. H., & Vlassov, V. (2020). Convjssp: Convolutional learning for
job-shop scheduling problems. In 2020 19th IEEE International Conference on Machine Learning and Applications (ICMLA) (pp. 1483-1490).
Zeng, Y., Zhou, J., & Du, L. (2021). RLC-GNN: An improved deep architecture for spatial-
based graph neural network with application to fraud detection. Applied Sciences, 11(12), 5656.
Zhang, C., Song, W., Cao, Z., Zhang, J., Tan, P. S., & Xu, C. (2020). Learning to dispatch
for job shop scheduling via deep reinforcement learning. Advances in Neural Information Processing Systems, 33, 1621–1632.