Computational Modeling and Smart Optimization For Ballistic Protection: A Systematic Review of Materials, Structural Response, Energy Systems, Control, and Predictive Maintenance

Abstract

Computational modeling and intelligent optimization can be used to enhance ballistic protection, but results need to be carefully interpreted based on the evidence of validation. The applications of this systematic review to protective materials, structural response, model calibration, uncertainty assessment and post-impact damage detection are discussed. Searches of Europe PMC and OpenAlex, with an existing exploratory search register, were performed on 30 October 2026 for publications first available from January 2015 to October 2026. Using PRISMA 2020 reporting, 341 different records were screened, 16 full reports were assessed and 14 studies were included. Structured narrative synthesis was conducted; meta-analysis was not performed because applications, reference data, validation procedures, and outcome measures were heterogeneous. The approaches included were finite element or molecular simulations in conjunction with neural networks, Gaussian processes, evolutionary optimization, inverse identification, and uncertainty quantification. The mean indentation calibration error was decreased from 17.27% to 1.74%, the damage-segmentation F1-scores were in the range of 87.39–92.74%, and the computational screening workflow only required approximately 5.16% of the estimated exhaustive finite element (EFE) runtime. These results are different computational tasks and cannot provide a common measure of protective reliability. Independent physical validation, uncertainty characterization and reproducibility was limited. Independent physical benchmarks, separating calibration from evaluation, explicit model-discrepancy assessment, and accessible data and code are necessary for future progress, and must be considered as part of the steps involved.