Topology Optimization of Brake Disc Inner Carrier and Development of a Parametric Predictive Model Using Machine Learning

Reducing unsprung mass in high-performance braking systems is critical for vehicle dynamics, but applying topology optimization (TO) to conventional solid discs under combined thermo-mechanical loading often causes “thermal entrapment” and structural failure. This study addresses this limitation by proposing a floating disc architecture and a two-stage computational framework. First, TO via the SIMP algorithm was applied to the Aluminum 7075-T6 inner carrier strictly under mechanical loads, decoupling artificial thermal stresses and yielding a fixed, optimized geometry. Results show the TO achieved a drastic 53% mass reduction (from 0.218 kg to 0.102 kg) specifically for the inner carrier component. Second, a high-fidelity Surrogate Model (Digital Twin) of this new geometry was developed to bypass the immense computational cost of multi-parameter, non-linear thermo-mechanical Finite Element Analysis (FEA). Utilizing Latin Hypercube Sampling (LHS) across 15 design scenarios and the Genetic Aggregation algorithm, the surrogate model was trained to predict real-time responses. Subsequently, the Digital Twin predicted stress and temperature fields with near-perfect accuracy (R² ≈ 0.9998), rapidly identifying the limit braking scenario. Fatigue analysis confirmed the final component safely withstands 10⁶ extreme braking cycles (safety factor 1.65).