Accelerated High-Power Blue Laser Design Cycle Enabled by Deep Neural Networks

The US Navy needs a blue laser system (BLS) delivering high-peak-power pulses with high-repetition rate for standoff oceanographic sensing applications from aircraft. Current state-of-the-art (SOTA) blue lasers meet some of the required characteristics, but none can simultaneously meet all. This is in-part due to the complexity of the BLS architecture, which requires a time-consuming iterative process between experiments and design optimization to maximize device performance. Due to the large number of variables involved, such a conventional systematic study is impractical as it stresses the cost and timeline for laser development. To overcome these challenges, Aqwest is now developing an automated laser design process using neural networks (NN) and machine learning (ML) algorithms.Ā The NN-ML process is an emerging powerful alternative to the conventional optimization. This data-driven approach offers to replace much of the computationally taxing multi-physics (MP) simulations in optimization loops, and allows for loop automation to assist the design process and reduce design time. The NN-ML process offers generating BLS designs meeting the Navy size, weight, performance, and reliability requirements orders of magnitude faster compared to the conventional design process. This project will develop and demonstrate fully automated BLS design algorithms using the NN-ML methodology. A power-scalable BLS prototype meeting the NavyÆs size, weight, and performance will be created for experimental verification of the automated designs. One objective is to demonstrate that the BLS performance metrics are met with less than +/- 5% variations from the target performance specifications. Another objective is to attain design process acceleration by a factor of 50 compared to the conventional ōmanualö laser design method. Fully automated BLS design algorithms with detailed user manual and documentations will be delivered to the Navy.