Modular Learning and Optimization for Planning of Discrete Event Systems
Doctoral thesis, 2021

Optimization of industrial processes, such as manufacturing cells, can have great impact on their performance. Finding optimal solutions to these large-scale systems is, however, a complex problem. They typically include multiple subsystems, and the search space generally grows exponentially with each subsystem. This is usually referred to as the state explosion problem and is a well-known problem within the control and optimization of automation systems.

This thesis proposes two main contributions to improve and to simplify the optimization of these systems. The first is a new method of solving these optimization problems using a compositional optimization approach. This integrates optimization with techniques from compositional supervisory control using modular formal models, dividing the optimization of subsystems into separate subproblems. The second is a modular learning approach that alleviates the need for prior knowledge of the systems and system experts when applying compositional optimization.

The key to both techniques is the division of the large system into smaller subsystems and the identification of local behavior in these subsystems, i.e. behavior that is independent of all other subsystems. It is proven in this thesis that this local behavior can be partially optimized individually without affecting the global optimal solution. This is used to reduce the state space in each subsystem, and to construct the global optimal solution compositionally.

The thesis also shows that the proposed techniques can be integrated to compute global optimal solutions to large-scale optimization problems, too big to solve based on traditional monolithic models.

discrete optimization

large-scale optimization

modular learning

automation

discrete event systems

active learning

Compositional optimization

ED-salen, Hörsalsvägen 11
Opponent: Professor Spiridon Reveliotis, Georgia Tech, USA

Author

Fredrik Hagebring

Chalmers, Electrical Engineering, Systems and control

Compositional Optimization of Discrete Event Systems

IEEE International Conference on Automation Science and Engineering,; Vol. 2018-August(2018)p. 849-856

Paper in proceeding

Time-optimal control of large-scale systems of systems using compositional optimization

Discrete Event Dynamic Systems: Theory and Applications,; Vol. 29(2019)p. 411-443

Journal article

Active Learning of Modular Plant Models

IFAC-PapersOnLine,; Vol. 53(2020)p. 296-302

Paper in proceeding

Modular Supervisory Synthesis for Unknown Plant Models Using Active Learning

IFAC-PapersOnLine,; Vol. 53(2020)p. 324-330

Paper in proceeding

On Optimization of Automation Systems: Integrating Modular Learning and Optimization

IEEE Transactions on Automation Science and Engineering,; Vol. In Press(2022)

Journal article

Automation systems are found everywhere today; within manufacturing, logistics, communication, and countless other areas. Furthermore, these systems are becoming increasingly complex and often include multiple subsystems; like machines, vehicles, etc. Optimization is often used to maximize their performance, but deriving the best sequence of operations is a difficult task that requires a high level of expertise and system knowledge. Furthermore, the systems are often so large that they can’t be easily scheduled even with the most powerful computers. This thesis presents two methods that address these problems.

Compositional Optimization is specifically designed for the optimization of large-scale automation systems. The strength comes from the ability to divide the large problem into smaller chunks, processing each subsystem individually. While this allows the method to optimize much larger systems, it still requires a high level of expertise due to a specific type of optimization model.

Modular Active Learning learns the optimization models for Compositional Optimization automatically from a simulation of the system. The user can input structural information, such as the number of subsystems and which operations that affect each subsystem. The more structural information that is added, the more efficient the learning and subsequent optimization will be.

Together, these two methods allow large-scale automation systems to be optimized with a minimum level of expertise and system knowledge.

Subject Categories

Production Engineering, Human Work Science and Ergonomics

Robotics

Areas of Advance

Transport

Production

ISBN

978-91-7905-555-4

Doktorsavhandlingar vid Chalmers tekniska högskola. Ny serie: 5022

Publisher

Chalmers

ED-salen, Hörsalsvägen 11

Online

Opponent: Professor Spiridon Reveliotis, Georgia Tech, USA

More information

Latest update

11/13/2023