Enhanced predictive modeling framework for multi-objective global optimization of passenger car rear seat using...

Area of Science:

• Automotive Engineering
• Computational Mechanics
• Optimization Theory

Background:

• Single approximation models struggle with highly nonlinear data in automotive seat design.
• Accurate fitting of complex feature data is crucial for effective multi-objective optimization.
• Existing methods may not fully capture the nonlinear responses under dynamic conditions like seat crashes.

Purpose of the Study:

• To propose a Hybrid Approximation Models based on the Multi-Species Approximation Model (HAM-MSAM) for high fitting accuracy.
• To introduce a HAM-MSAM-based Approximation-Based Global Multi-Objective Optimization Design (ABGMOOD) strategy.
• To optimize the multi-objective design of a passenger car's rear seat.

Main Methods:

• Constructed HAM-MSAM using an experimentally validated finite element model and an experimental design-generated training set.
• Employed the HAM-MSAM-based ABGMOOD strategy for rear seat optimization.
• Validated HAM-MSAM's ability to capture nonlinear responses under crash conditions against existing hybrid models.
• Compared ABGMOOD results with a classical local multi-objective optimization strategy.

Main Results:

• HAM-MSAM demonstrated superior fitting accuracy for nonlinear seat crash data compared to existing methods.
• The ABGMOOD strategy achieved significant improvements in economy and weight reduction for the rear seat.
• Optimized rear seats met regulatory safety requirements, though slightly lower than the local optimization scheme.
• The optimized design showed notable enhancements in safety, economy, and weight reduction.

Conclusions:

• The HAM-MSAM-based ABGMOOD strategy is feasible and effective for automotive seat multi-objective optimization.
• This approach offers substantial advantages in economy and weight reduction.
• The optimized rear seat design balances performance, safety, and efficiency, providing valuable insights for future engineering challenges.