.png){kind=icon}
.png){kind=icon}
Return
Optimization Design of Vibration Characteristics of New Energy Lithium Batteries Based on SCSO-BP Machine Learning Based on Pareto Classification and Genetic Algorithm
.png){kind=icon}
PRE
AI
Y
N
L
DOI：10.1007/s40799-025-00840-2.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This study introduces an integrated methodological framework to address critical mechanical challenges in ternary lithium battery enclosures. The approach initiates with a variable-density topology optimization platform implemented through Minimalist GNU for Windows 64-bit (MingW-W64), incorporating Latin hypercube sampling for efficient design space exploration. A meticulously calibrated non-dominated sorting genetic algorithm (NSGA-II) performs multi-objective optimization, simultaneously minimizing von-mises stress concentration while maximizing structural stiffness as competing design criteria. The framework employs Gaussian process-based Kriging interpolation to construct high-fidelity response surfaces, facilitating the identification of Pareto-optimal configurations. Subsequently, a 60-parameter back propagation (BP) neural network optimized via Sand Cat Swarm Optimization (SCSO) demonstrates exceptional predictive accuracy. Comprehensive dynamic validation incorporates triaxial random vibration testing (X/Y/Z axes per ISO 19453-6:2020 standards), with accompanying finite element analysis quantifying key performance metrics including stress amplification factors (SAF < 2.5), displacement spectra, and component-level safety margins. Comparative simulation results reveal statistically significant improvements in the optimized design, including reduced stress concentrations, enhanced natural frequency characteristics, and superior stiffness-to-mass ratios relative to baseline configurations. The final design maintains full compliance with critical electrical safety requirements while demonstrating robust reliability for industrial applications.
Keywords:
Pure electric vehicle
Multi-objective genetic algorithm
SCSO-BP optimization
Journal
E
IF:
1.9
Papers:
50
Citations:
1.6K
Organization
Cited Papers