"Deep reinforcement learning for engineering design through topology optimization of elementally discretized design domains". (June 2022)

• Record Type: Journal Article
• Authors: Brown, Nathan K.; Garland, Anthony P.; Fadel, Georges M.; Li, Gang
• Is Part Of: Materials & design. Volume 218(2022)
• Publication Date: 2022-06
• Subjects: Reinforcement learning -- Topology optimization -- Deep learning -- Engineering design -- Structural design -- Data-driven; Materials -- Periodicals; Engineering design -- Periodicals; Matériaux -- Périodiques; Conception technique -- Périodiques; Electronic journals; 620.11
• DOI: 10.1016/j.matdes.2022.110672
• Languages: English
• ISSNs: 0264-1275

Abstract

Graphical abstract: Highlights: Deep reinforcement learning was used to find the optimal sequences of design decisions to optimize elementally discretized structures. This work offers a gradient-free approach to solving the compliance minimization topology optimization problem. The deep reinforcement learning agent was trained using an interactive environment built to capture compliance minimization relevant information. The agent designed optimal topologies using a two-step progressive refinement method leading to improved efficiency and solution detail. Abstract: Advances in machine learning algorithms and increased computational efficiencies give engineers new capabilities and tools to apply to engineering design. Machine learning models can approximate complex functions and, therefore, can be useful for various tasks in the engineering design workflow. This paper investigates using reinforcement learning (RL), a subset of machine learning that teaches an agent to complete a task through accumulating experiences in an interactive environment, to automate the designing of 2D discretized topologies. RL agents use past experiences to learn sequential sets of actions to best achieve some objective. In the proposed environment, an RL agent can make sequential decisions to design a topology by removing elements to best satisfy compliance minimization objectives. After each action, the agent receives feedback by evaluating how well the current topology satisfies the design objectives. After training, the agent was tasked with designing optimal topologies under various load cases. The agent's proposed designs had similar or better compliance minimization performance to those produced by traditional gradient-based topology optimization methods. These results show that a deep RL agent can learn generalized design strategies to satisfy multi-objective design tasks and, therefore, shows promise as a tool for arbitrarily complex design problems across many domains.