Cluster analysis and neural network-based metamodeling of priority rules for dynamic sequencing

Abstract

Most sequencing problems deal with deterministic environments where all information is known in advance. However, in real-world problems multiple sources of uncertainty need to be taken into consideration. To model such a situation, in this article, a dynamic sequencing problem with random arrivals, processing times and due-dates is considered. The examined system is a manufacturing line with multiple job classes and sequence-dependent setups. The performance of the system is measured under the metrics of mean WIP, mean cycle time, mean earliness, mean tardiness, mean absolute lateness, and mean percentage of tardy jobs. Twelve job dispatching rules for solving this problem are considered and evaluated via simulation experiments. A statistically rigorous analysis of the solution approaches is carried out with the use of unsupervised and supervised learning methods. The cluster analysis of the experimental results identified classes of priority rules based on their observed performance. The characteristics of each priority rule class are documented and areas in objective space not covered by existing rules are identified. The functional relationship between sequencing priority rules and performance metrics of the production system was approximated by artificial neural networks. Apart from gaining insight into the mechanics of the sequencing approaches the results of this article can be used (1) as a component for prediction systems of dispatching rule output, (2) as a guideline for building new dispatching heuristic with entirely different characteris-B A. S. Xanthopoulos tics than existing ones, (3) to significantly decrease the length of what-if simulation studies.

Review of job shop scheduling research and its new perspectives under Industry 4.0

“…(2010) [52] put forward a hybrid method based on variable neighborhood search(VNS) and artificial neural network (ANN) for dynamic job shop scheduling to deal with random job arrivals and machine breakdowns. Xanthopoulos and Koulouriotis (2015) [53] applied BP neural networks to approximate the functional relationship between dynamic sequencing priority rules and performance metrics of the production system. The results of the trained BP neural networks for scheduling can be used to predict outputs of dispatching rule systems, direct to build new dispatching heuristic and significantly decrease the time of simulation studies.…”

Analysis of an individual‐based influenza epidemic model using random forest metamodels and adaptive sequential sampling

“…Therefore, many supervised learning algorithms can be effectively used to develop simulation metamodels. These include neural networks (Alam, McNaught, & Ringrose, 2004; Can & Heavey, 2012; Kuo, Yang, Peters, & Chang, 2007; Xanthopoulos & Koulouriotis, 2018), kriging/Gaussian process (Dancik, Jones, & Dorman, 2010; Dosi, Pereira, & Virgillito, 2018; Kleijnen, 2009; Salle & Yıldızoğlu, 2014), SVR (Clarke, Griebsch, & Simpson, 2005; Edali & Yücel, 2018; Fonoberova, Fonoberov, & Mezić, 2013; Zhou, Shao, Jiang, Zhou, & Shu, 2015), RFs (Edali & Yücel, 2019; Villa‐Vialaneix, Follador, Ratto, & Leip, 2012), multivariate adaptive regression splines (Bozağaç, Batmaz, & Oğuztüzün, 2016; Friedman, 1991), radial basis functions (Hussain, Barton, & Joshi, 2002; Jakobsson, Patriksson, Rudholm, & Wojciechowski, 2010; Mullur & Messac, 2006) and first‐ and second‐order linear regression models (Durieux & Pierreval, 2004; Grow, 2017; Happe, Kellermann, & Balmann, 2006; Kleijnen & Deflandre, 2006). There are several studies in the literature which compare different subsets of these techniques based on different criteria such as accuracy, robustness, interpretability and efficiency (i.e., runtime) (Clarke, Griebsch, & Simpson, 2005; Li, Ng, Xie, & Goh, 2010; Østergård, Jensen, & Maagaard, 2018; Van Gelder, Das, Janssen, & Roels, 2014; Villa‐Vialaneix, Follador, Ratto, & Leip, 2012).…”

Gated‐Attention Model with Reinforcement Learning for Solving Dynamic Job Shop Scheduling Problem

“…However, using mathematical optimization to solve production scheduling problems is only effective in smaller and less complex environments. Approximate algorithms which include local search algorithms [8], meta‐heuristics [9] and artificial intelligence [10] on the other hand, have been widely used to deal with various scheduling tasks. These solution approaches, however, are successful in producing optimal or near‐optimal solutions in static contexts [11], rendering them unsuitable for scheduling large scale and dynamic real‐world production problems.…”