Intelligent design exploration method for complex engineered system architecture generation

ABSTRACT

Owing to the discontinuous specificity and complexity of architecture design space, the issue of selecting and combining components comprising the engineered system, numerous constraints and associations need to be accounted for, adding up to a complex and substantial cognitive load on the system architects, which makes it challenging to tackle the current demand of adaptive improvements or innovative upgrading of the existing mature architectural solutions. To this end, this paper proposes an intelligent design exploration method for complex system architecture generation with reinforcement learning. The architectural design space (ADS) is identified by defining the dimensions of ADS, including model, quantity, and design chain, as well as the mathematical boundaries and representation to facilitate computable intelligent design exploration. On this basis, by adopting AI techniques primarily based on reinforcement learning, a massive and reliable architectural scheme is rapidly generated, and a more satisfying and robust architectural solution is selected by accessing the fuzzy Pareto frontier. Validation of the method is demonstrated through a case study of a launch vehicle's first and second-stage separation system. This research contributes valuable insights to overcoming the limitations of traditional techniques and enhancing the efficiency of the generative design and decision-making for complex engineered system architecture.

KEYWORDS:

• Architecture generation
• intelligent design exploration
• reinforcement learning
• architectural design space

Acknowledgments

This paper is an outcome of the International Systems Realisation Partnership between the Institute for Industrial Engineering @ The Beijing Institute of Technology, China, The Systems Realisation Laboratory @ The University of Oklahoma, USA, the Design Engineering Laboratory @ Purdue, USA, and The Systems Realisation Laboratory @ The University of Liverpool, UK. Thanks for the support of the Key Laboratory of Industry Knowledge & Data Fusion Technology and Application, Ministry of Industry and Information Technology, Beijing Institute of Technology.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the National Natural Science Foundation of China (NSFC 52105241), Beijing Institute of Technology Research Fund Program for Young Scholars, and also financial support from the National Ministries Projects of China (50923010101, D020101).