https://github.com/kmoraza/bo-based-damper_placement_optimization--approach2
Optimal damper placement in automotive mechanical suspension implemented using Bayesian Optimization (BO) and written in Java
https://github.com/kmoraza/bo-based-damper_placement_optimization--approach2
aerodynamics automotive-suspension bayesian-optimization mathematics mechanical-engineering optimization-algorithms optimization-methods optimization-techniques pareto-front physics random-forest simulated-annealing
Last synced: about 1 month ago
JSON representation
Optimal damper placement in automotive mechanical suspension implemented using Bayesian Optimization (BO) and written in Java
- Host: GitHub
- URL: https://github.com/kmoraza/bo-based-damper_placement_optimization--approach2
- Owner: KMORaza
- Created: 2025-04-26T22:23:40.000Z (6 months ago)
- Default Branch: main
- Last Pushed: 2025-04-27T09:56:03.000Z (6 months ago)
- Last Synced: 2025-05-29T02:07:55.076Z (4 months ago)
- Topics: aerodynamics, automotive-suspension, bayesian-optimization, mathematics, mechanical-engineering, optimization-algorithms, optimization-methods, optimization-techniques, pareto-front, physics, random-forest, simulated-annealing
- Language: Java
- Homepage: https://bo-optimal-damper-placement-2.netlify.app/
- Size: 48.8 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
## Optimale Dämpferplatzierung in der mechanischen Fahrzeugaufhängung, entwickelt mit Bayesscher Optimierung und geschrieben in der Java-Programmiersprache (Optimal damper placement in automotive mechanical suspension implemented using Bayesian Optimization and written in Java)
A multi-objective solution that utilizes Bayesian Optimization (BO) to optimize damper placement and parameters in an automotive suspension system.
The solution is to minimize three objectives—comfort (ISO 2631), vibration (frequency-weighted PSD), and handling (tire load variation)—across different road profiles (urban, highway, off-road) and damping profiles (linear, digressive, progressive).
* Defines fixed vehicle parameters (e.g., sprung mass = 1500 kg, unsprung mass = 50 kg per wheel, tire stiffness = 200000 N/m).
* Implements a Gaussian Process (GP) regression model & uses a Radial Basis Function (RBF) kernel
* Supports incremental updates to the kernel matrix using the Sherman-Morrison formula for efficiency.
* Provision for mean predictions and for uncertainty estimates, crucial for Bayesian optimization’s exploration-exploitation trade-off.
* Uses three surrogate ensemble instances (one per objective: comfort, vibration, handling) to model the objective function.
* Initializes with 30 random points, then iteratively selects new points using Simulated Annealing to maximize an active learning score combining Expected Hypervolume Improvement (EHVI), variance, and diversity.
* Maintains Pareto front to store non-dominated solutions, with adaptive weights based on crowding entropy to balance objectives.
* Uses EHVI to balance comfort, vibration, and handling, ensuring a diverse Pareto front.
* The ensemble of GP, RF, and NN provides robust predictions and uncertainty estimates, critical for Bayesian optimization.
* Incremental updates (e.g., Sherman-Morrison in GP) improve efficiency as new points are added.
* Parallel evaluation of objectives reduces runtime for expensive simulations.
* Simulated Annealing optimizes the acquisition function, balancing exploration and exploitation.
* Handles invalid objectives (NaN, infinite, or excessively large) by retrying with perturbed parameters or assigning high penalties.
---
Check [__*webpage*__](https://bo-optimal-damper-placement-2.netlify.app/) about the solution.