An Adaptive Search Algorithm for Multiplicity Dynamic Flexible Job Shop Scheduling with New Order Arrivals

Symmetry, Journal Year: 2024, Volume and Issue: 16(6), P. 641 - 641

Published: May 22, 2024

In today’s customer-centric economy, the demand for personalized products has compelled corporations to develop manufacturing processes that are more flexible, efficient, and cost-effective. Flexible job shops offer organizations agility cost-efficiency traditional lack. However, dynamics of modern manufacturing, including machine breakdown new order arrivals, introduce unpredictability complexity. This study investigates multiplicity dynamic flexible shop scheduling problem (MDFJSP) with arrivals. To address this problem, we incorporate fluid model propose a randomized adaptive search (FRAS) algorithm, comprising construction phase local phase. Firstly, in phase, heuristic an online tracking policy generates high-quality initial solutions. Secondly, employ improved tabu procedure enhance efficiency solution space, incorporating symmetry considerations. The results numerical experiments demonstrate superior effectiveness FRAS algorithm solving MDFJSP when compared other algorithms. Specifically, proposed demonstrates quality relative existing algorithms, average improvement 29.90%; exhibits acceleration speed, increase 1.95%.

Language: Английский

---

Dynamic scheduling of hybrid flow shop problem with uncertain process time and flexible maintenance using NeuroEvolution of Augmenting Topologies

IET Collaborative Intelligent Manufacturing, Journal Year: 2024, Volume and Issue: 6(3)

Published: Sept. 1, 2024

Abstract A hybrid flow shop is pivotal in modern manufacturing systems, where various emergencies and disturbances occur within the smart context. Efficiently solving dynamic scheduling problem (HFSP), characterised by release times, uncertain job processing flexible machine maintenance has become a significant research focus. NeuroEvolution of Augmenting Topologies (NEAT) algorithm proposed to minimise maximum completion time. To improve NEAT algorithm's efficiency effectiveness, several features were integrated: multi‐agent system with autonomous interaction centralised training develop parallel policy, maintenance‐related action for optimal decision learning, proactive avoid waiting jobs at moments, thereby exploring broader solution space. The performance trained model was experimentally compared Deep Q‐Network (DQN) five classical priority dispatching rules (PDRs) across scales. results show that achieves better solutions responds more quickly changes than DQN PDRs. Furthermore, generalisation test demonstrate NEAT's rapid problem‐solving ability on instances different from set.

Language: Английский

---

Dynamic Scheduling for Large-Scale Flexible Job Shop Based on Noisy DDQN

International Journal of Network Dynamics and Intelligence, Journal Year: 2023, Volume and Issue: unknown, P. 100015 - 100015

Published: Dec. 21, 2023

Article Dynamic Scheduling for Large-Scale Flexible Job Shop Based on Noisy DDQN Tingjuan Zheng 1,2, Yongbing Zhou 1, Mingzhu Hu and Jian Zhang 1,* 1 Institute of Advanced Design Manufacturing, School Mechanical Engineering Southwest Jiaotong University, Chengdu 610031, China 2 Guizhou Aerospace Electric Co., Ltd., Guiyang 550009, * Correspondence: [email protected] Received: 3 July 2023 Accepted: 8 October Published: 21 December Abstract: The large-scale flexible job shop dynamic scheduling problem (LSFJSDSP) has a more complex solution space than the original because increase in number jobs machines, which makes traditional algorithm unable to meet actual production requirements terms quality time. To address this problem, we develop model based noisynet-double deep Q-networks (N-DDQNs), takes minimum expected completion time as optimization objective thoroughly into account two factors (the new arrival stochastic processing time). Firstly, Markov decision process is constructed workshop, corresponding reasonable state space, action reward function are designed. problems (of stability unsatisfactory strategy selection) conventional exploration method DDQNs, learnable noise parameters added DDQNs create N-DDQN framework, where uncertainty weight added. Secondly, form realize automatic exploration. Hence, issue solved that may result selection. proposed method, significant flexibility efficacy, demonstrated (by experimental findings) be superior compound rules tackling problems.

Language: Английский

---

Genetic Programming for Dynamic Flexible Job Shop Scheduling: Evolution With Single Individuals and Ensembles

IEEE Transactions on Evolutionary Computation, Journal Year: 2023, Volume and Issue: 28(6), P. 1761 - 1775

Published: Nov. 21, 2023

Dynamic flexible job shop scheduling is an important but difficult combinatorial optimisation problem that has numerous real-world applications. Genetic programming been widely used to evolve heuristics solve this problem. Ensemble methods have shown promising performance in many machine learning tasks, previous attempts combine genetic with ensemble techniques are still limited and require further exploration. This paper proposes a novel method uses population consisting of both single individuals ensembles. The main contributions include: 1) developing evolves comprising ensembles, allowing breeding between them explore the search space more effectively; 2) proposing construction selection strategy form ensembles by selecting diverse complementary individuals; 3) designing new crossover mutation operators produce offspring from Experimental results demonstrate proposed outperforms existing traditional most scenarios. Further analyses find success attributed enhanced diversity extensive exploration achieved method.

Language: Английский

---

Dynamic scheduling for multi-objective flexible job shop via deep reinforcement learning

Applied Soft Computing, Journal Year: 2025, Volume and Issue: 171, P. 112787 - 112787

Published: Jan. 25, 2025

Language: Английский

---