Michael Hellwig

h-index11
2papers
363citations

2 Papers

5.8NEMay 5
A Benchmarking Suite for Flexible Job Shop Scheduling Problems with Worker Flexibility under Uncertainty

David Hutter, Thomas Steinberger, Michael Hellwig

This paper addresses the Flexible Job Shop Scheduling Problem and its extension with Worker Flexibility, which integrates workforce assignment into machine-operation scheduling. Diverse solvers have been proposed across multiple optimization domains including Mathematical Programming, Constraint Programming, and Simulation-Based Optimization, or Simulation-based Optimization. These are often tailored to narrow use cases and validated on limited test problem sets, hindering cross-domain comparison. To overcome this, a comprehensive benchmarking environment built on 402 standardized Flexible Job Shop Scheduling Problem instances is introduced and systematically extended to include worker flexibility. This creates a hitherto unique collection of ready-to-use worker flexibility instances. The benchmark suite features several metrics for algorithm performance assessment, the visualization of algorithmic results, as well as state-of-the-art baseline results. This enables rigorous, reproducible, and comparable performance analysis between solvers and scheduling problem subdomains. Through the simulation-based integration of uncertainties in processing times as well as resource availabilities, the environment supports the development and evaluation of robust optimization strategies. The present work lays a foundation for targeted algorithm development and consistent performance evaluation in production scheduling research.

1.8NEJul 26, 2018Code
A Linear Constrained Optimization Benchmark For Probabilistic Search Algorithms: The Rotated Klee-Minty Problem

Michael Hellwig, Hans-Georg Beyer

The development, assessment, and comparison of randomized search algorithms heavily rely on benchmarking. Regarding the domain of constrained optimization, the number of currently available benchmark environments bears no relation to the number of distinct problem features. The present paper advances a proposal of a scalable linear constrained optimization problem that is suitable for benchmarking Evolutionary Algorithms. By comparing two recent EA variants, the linear benchmarking environment is demonstrated.