Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Sample-Efficient Bayesian Transfer Learning for Online Machine Parameter Optimization

Philipp Wagner, Tobias Nagel, Philipp Leube, Marco F. Huber

TL;DR

This work introduces a method to optimize the machine parameters in the system itself using a Bayesian optimization (BO) algorithm, and uses a transfer learning approach in order to identify an optimum with minimal iterations, resulting in a cost-effective transfer learning algorithm.

Abstract

Correctly setting the parameters of a production machine is essential to improve product quality, increase efficiency, and reduce production costs while also supporting sustainability goals. Identifying optimal parameters involves an iterative process of producing an object and evaluating its quality. Minimizing the number of iterations is, therefore, desirable to reduce the costs associated with unsuccessful attempts. This work introduces a method to optimize the machine parameters in the system itself using a Bayesian optimization algorithm. By leveraging existing machine data, we use a transfer learning approach in order to identify an optimum with minimal iterations, resulting in a cost-effective transfer learning algorithm. We validate our approach on a laser machine for cutting sheet metal in the real world.

Sample-Efficient Bayesian Transfer Learning for Online Machine Parameter Optimization

TL;DR

This work introduces a method to optimize the machine parameters in the system itself using a Bayesian optimization (BO) algorithm, and uses a transfer learning approach in order to identify an optimum with minimal iterations, resulting in a cost-effective transfer learning algorithm.

Abstract

Correctly setting the parameters of a production machine is essential to improve product quality, increase efficiency, and reduce production costs while also supporting sustainability goals. Identifying optimal parameters involves an iterative process of producing an object and evaluating its quality. Minimizing the number of iterations is, therefore, desirable to reduce the costs associated with unsuccessful attempts. This work introduces a method to optimize the machine parameters in the system itself using a Bayesian optimization algorithm. By leveraging existing machine data, we use a transfer learning approach in order to identify an optimum with minimal iterations, resulting in a cost-effective transfer learning algorithm. We validate our approach on a laser machine for cutting sheet metal in the real world.

Paper Structure

This paper contains 11 sections, 15 equations, 3 figures, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Problem Formulation
  4. Transfer Learning for Bayesian Optimization
  5. Gaussian Process
  6. General Description
  7. Evaluation
  8. Data Acquisition
  9. Thermal Cutting Simulation
  10. Real Hardware Experiment
  11. Conclusion and Future Work

Figures (3)

  • Figure 1: Demonstration of our machine parameter optimization in thermal cutting with six iterations. Iteratively adjusting the machine parameters reduces the burr from initially high (left) to barely visible (right).
  • Figure 2: Minimal burr height predicted after each BO iteration as well as the corresponding standard deviation for every method averaged over ten trials. To improve readability, we shift the data slightly on the x-axis for each method. The optimization trials are performed on a surrogate model.
  • Figure 3: Resulting burr height measured after every optimization step for one optimization run. The optimization was performed on an 8mm steel sheet and a power of 6kW. Dashed lines indicate the minimal burr reached for each method after six iterations. Missing values indicate cut interruptions.