Multi-objective optimization of a SWaP-refined miniature Stirling cryocooler using an integrated ANN-Genetic algorithm-based decision-making approach.

Miniature Stirling cryocoolers are vital for modern high-temperature Infrared (IR) detectors due to their precise cooling and SWaP (Size, Weight, and Power) compliance. This study proposes an integrated ANN-based multiobjective genetic algorithm (MOGA) optimization model with technique for order preference by similarity to ideal solution (TOPSIS) decision-making approach to optimize the coefficient of performance (COP) and work input (Winput) of a SWaP-refined Stirling cryocooler. According to the 2nd-order semi-adiabatic thermodynamics model (SATM) analysis, the design parameters, including motor rotational speed (RPM), phase angle between the piston and displacer (ϕ), working fluid initial charging pressure (Po), cooling temperature (Te), and piston stoke (Xp), have a significant impact on the COP and Winput of the SWaP-refined cryocooler, and therefore, optimization is required. Using a dataset from the SATM model based on SWaP constraints, a MOGA integrated with an ANN-trained model is applied to optimize these design parameters for the maximum COP and minimum Winput. A 100 optimized solutions were generated using the MOGA Pareto front, with the best optimized solution identified through the TOPSIS decision-making approach. At the TOPSIS closeness coefficient of Ci = 0.741, the relative errors between the ANN-MOGA results and the semi-adiabatic thermodynamic model are 2.9% for COP and 5% for Winput. With overall prediction errors less than 5%, the proposed integrated ANN-based MOGA optimization model offers an efficient and reliable approach for optimizing the design parameters and performance of SWaP-configured miniature stirling cryocoolers.