Jens Heger

Towards improved dispatching rules for complex shop floor scenarios: a genetic programming approach

Developing dispatching rules for manufacturing systems is a process, which is time- and cost-consuming. Since there is no good general rule for different scenarios and objectives automatic rule search mechanism are investigated. In this paper an approach using Genetic Programming (GP) is presented. The priority rules generated by GP are evaluated on dynamic job shop scenarios from literature and compared with manually developed rules yielding very promising results also interesting for Simulation Optimization in general.

Generating dispatching rules for semiconductor manufacturing to minimize weighted tardiness

Dispatching rules play an important role especially in semiconductor manufacturing scheduling, because these fabrication facilities are characterized by high complexity and dynamics. The process of developing and adapting dispatching rules is currently a tedious, largely manual task. Coupling Genetic Programming (GP), a global optimization meta-heuristic from the family of Evolutionary Algorithms, with a stochastic discrete event simulation of a complex manufacturing system we are able to automatically generate dispatching rules for a scenario from semiconductor manufacturing. Evolved dispatching rules clearly outperform manually developed rules from literature.

Dynamic adjustment of dispatching rule parameters in flow shops with sequence-dependent set-up times

Decentralised scheduling with dispatching rules is applied in many fields of production and logistics, especially in highly complex manufacturing systems. Since dispatching rules are restricted to their local information horizon, there is no rule that outperforms other rules across various objectives, scenarios and system conditions. In this paper, we present an approach to dynamically adjust the parameters of a dispatching rule depending on the current system conditions. The influence of different parameter settings of the chosen rule on the system performance is estimated by a machine learning method, whose learning data is generated by preliminary simulation runs. Using a dynamic flow shop scenario with sequence-dependent set-up times, we demonstrate that our approach is capable of significantly reducing the mean tardiness of jobs.

Gaussian Processes for Dispatching Rule Selection in Production Scheduling: Comparison of Learning Techniques

Decentralized scheduling with dispatching rules is applied in many fields of logistics and production, especially in semiconductor manufacturing, which is characterized by high complexity and dynamics. Many dispatching rules have been found, which perform well on different scenarios, however no rule has been found, which outperforms other rules across various objectives. To tackle this drawback, approaches, which select dispatching rules depending on the current system conditions, have been proposed. Most of these use learning techniques to switch between rules regarding the current system status. Since the study of Rasmussen [1] has shown that Gaussian processes as a machine learning technique have outperformed other techniques like neural networks under certain conditions, we propose to use them for the selection of dispatching rules in dynamic scenarios. Our analysis has shown that Gaussian processes perform very well in this field of application. Additionally, we showed that the prediction quality Gaussian processes provide could be used successfully.

Integration of demand forecasts in ABC‐XYZ analysis: practical investigation at an industrial company

Purpose – Item classification based on ABC‐XYZ analysis is of high importance for strategic supply and inventory control. It is common to perform the analysis with past consumption data. In this context, the purpose of this study is to test the hypothesis that an integration of demand forecasts can improve the performance of item classification, in particular the performance of ABC‐XYZ analysis.Design/methodology/approach – For the study, real data of an industrial enterprise in the mechanical engineering sector (focal company) were analyzed and evaluated.Findings – The study shows that a comprehensive data analysis of the focal company can recommend a specific implementation of the ABC‐XYZ classification. In contrast to the classic method of making the ABC‐XYZ analysis based on consumption data only, the approach developed in this paper offers considerable advantages. These are quantifiable in respect to an assumed optimal reference classification.Originality/value – The evaluation of the results is very...

Dispatching rule selection with Gaussian processes

Decentralized scheduling with dispatching rules is applied in many fields of logistics and production, especially in highly complex and dynamic scenarios, such as semiconductor manufacturing. Many dispatching rules are proposed in the literature, which perform well on specific scenarios. No rule is known, however, consistently outperforming all other rules. One approach to meet this challenge and improve scheduling performance is to select and switch dispatching rules depending on current system conditions. For this task machine learning techniques (e.g., Artificial Neural Networks) are frequently used. In this paper we investigate the use of a machine learning technique not applied to this task before: Gaussian process regression. Our analysis shows that Gaussian processes predict dispatching rule performance better than Neural Networks in most settings. Additionally, already a single Gaussian Process model can easily provide a measure of prediction quality. This is in contrast to many other machine learning techniques. We show how to use this measure to dynamically add additional training data and incrementally improve the model where necessary. Results therefore suggest, Gaussian processes are a very promising technique, which can lead to better scheduling performance (e.g., reduced mean tardiness) compared to other techniques.

Switching Dispatching Rules with Gaussian Processes

Decentralized scheduling with dispatching rules is applied in many fields of production and logistics, especially in highly complex manufacturing systems, e.g. semiconductor manufacturing. Nevertheless, no dispatching rule outperforms other rules across various objectives, scenarios and system conditions. In this paper we present an approach to dynamically select the most suitable rule for the current system conditions in real time. We calculate Gaussian process (GP) regression models to estimate each rule’s performance and select the most promising one. The data needed to create these models is gained by a few preliminary simulation runs of the selected job shop scenario from the literature. The approach to use global information to create the Gaussian process models leads to better local decision at the machine level. Using a dynamic job shop scenario we demonstrate, that our approach is capable of significantly reducing the mean tardiness of jobs.

Analysis of Priority Rule-Based Scheduling in Dual-Resource-Constrained Shop-Floor Scenarios

A lot of research on scheduling manufacturing systems with priority rules has been done. Most studies, however, concentrate on simplified scenarios considering only one type of resource, usually machines. In this study priority rules are applied to a more realistic scenario, in which machines and operators are dual-constrained and have a re-entrant process flow. Interdependencies of priority rules are analyzed by long-term simulation. Strength and weaknesses of various priority rule combinations are determined at different utilization levels. Further insights are gained by additionally solving static instances optimally by using a mixed integer linear program (MILP) of the production system and comparing the results with those of the priority rules.

Supporting non-hierarchical supply chain networks in the electronics industry

The European electronics industry faces a strong competition with far eastern and US manufacturers. They have to respond with improved flexibility to changing requirements and collaborate across the supply chain effectively capitalizing on collaborative decision making. On operational levels a lot of concepts and tools have been implemented, which support automatic information exchange to optimize the supply chains. However, the collaboration on tactical and strategic levels is not supported; especially new forming non-hierarchical networks are concerned. This paper describes the problem analysis as a basis for the project approach and first findings gained from four industrial use cases. The goal of the CONVERGE project is to fill the existing gap regarding collaboration on tactic and strategic level by providing a framework and tools for exchanging tactical and strategic information.

Optimal Scheduling for Automated Guided Vehicles (AGV) in Blocking Job-Shops

Promising developments and further improvements of cyber physical logistics systems (CPLS) and automated guided vehicles (AGV) lead to broader application of such systems in production environments and smart factories. In this study a new mixed integer linear program (MILP) is presented for the scheduling of AGVs in a flexible reentrant job shop with blocking. Optimal solutions to small instances of the complex scheduling problem in a make-to-order production, minimizing the make span, are calculated. Different numbers of jobs are considered. Feasible schedules for the machines and the AGVs are generated from different sized instances to evaluate the limits of the mathematical model.