b'Artificial intelligence- Artificial intelligence-based process control and optimization during directed based process controlenergy deposition manufacturing transforms manufacturing operations.and optimization forT he objective of this project was to develop a novel artificial intelligence-based process control and optimization methodology for advanced manufacturing. advanced manufacturing Process-informed design, which can transform manufacturing operations, requires online process control and optimization. The existing hardware level controllers, such as proportional integral derivative controllers, follow prescribed system settings and do not have the capability to take actions based on the ever-changing process states. Industrial companies employ simulation tools combined with artificial intelligence techniques for an accelerated trial and error design process. Intelligent PROJECT NUMBER:advanced manufacturing minimizes human inputs to the optimization process but 22A1059-047FP relies on an automated process-level control mechanism to generate optimal design variables and adaptive system settings for improved end-product properties. This TOTAL APPROVED AMOUNT:project provided an online interaction mechanism in the process-informed design by $820,000 over 2 years developing the capability to intelligently control and optimize advanced manufacturing PRINCIPAL INVESTIGATOR:processes. To achieve this goal, artificial intelligence-based control algorithms were Dewen Yushu developed using deep reinforcement learning. To reduce the computational expense with advanced manufacturing models, reduced order models were developed using CO-INVESTIGATORS: operator learning techniques. An integrated online artificial intelligence training Asa Monson, INL capability was developed based on the MOOSE Stochastic Tools Module. INLs existing Fande Kong, INL directed energy deposition simulation model was validated against experimental Michael McMurtrey, INL measurements at different operation conditions and enhanced to be able to simulate Peter German, INL the materials thermal-mechanical behavior with improved accuracy and better Zonggen Yi, INL efficiency. Preliminary demonstration and validation of the developed process control Xu Wu, North Carolina State University and optimization methodology was completed with extensive directed energy deposition scans and measurements using INLs Laser Engineering Net Shaping system. Major scientific accomplishments from this project. (a) Machine learning (ML) capability was enabled within MOOSE, and its performance was examined. (b) Directed energy deposition (DED) model was improved and validated through extensive experimental measurements. (c) Reduced order models (ROMs) were developed based on the high-fidelity DED model using deep operator network and Fourier neural operator to enable quick and accurate predictions of the maximum temperature throughout the simulation. (d) Artificial-intelligence-based process control system is designed and implemented.104'