來源ITmedia AI+ (日本)•較早收集於 83m
住友橡膠與 Fujitsu 合作，利用 AI 加速輪胎 FEM 分析
#ai-surrogate-model#digital-twin#industrial-ai#simulationai-surrogate-model-for-tire-fem-analysissumitomo rubber industriesfujitsufem
💡了解 AI 代理模型如何將工業模擬時間縮短 9 倍，徹底改變傳統研發工作流程。
⚡ 30 秒速覽
有什麼變化
將輪胎 FEM 分析時間從 45 分鐘縮短至 5 分鐘。
為什麼重要
此項發展展示了 AI 代理模型在工業工程中的實際應用，取代了計算成本高昂的傳統模擬。它為透過 AI 驅動的數位孿生技術加速製造業研發週期樹立了先例。
下一步行動
研究如何使用代理模型（如物理資訊神經網路 PINNs）來取代您工程工作流程中繁重的數值模擬。
誰應關注：Researchers & Academics
關鍵要點
- •將輪胎 FEM 分析時間從 45 分鐘縮短至 5 分鐘。
- •成功處理包含 60 萬個元素的複雜模擬。
- •由住友橡膠工業與 Fujitsu 共同開發。
- •AI 代理模型技術實現了輪胎性能預測的快速迭代。
🧠 深度解析
背景與延伸：來自公開資料，非原文內容。引用 20 個來源。
🔑 增強重點摘要
- •The AI model leverages a Graph Neural Network (GNN) algorithm, trained using Sumitomo Rubber's tire design expertise and actual design data.
- •The technology achieved an average accuracy of 87.7% in predicting tire-to-road contact shape compared to traditional FEM analysis.
- •The AI surrogate model is optimized for Fujitsu's upcoming Arm-based FUJITSU-MONAKA processor, aiming for enhanced inference speed and power efficiency.
- •This initiative is a key component of Sumitomo Rubber's long-term digital transformation strategy, "R.I.S.E. 2035," for tire design and development.
- •Fujitsu plans to commercialize this AI inference platform, integrating FUJITSU-MONAKA and GNN, through its "Fujitsu Kozuchi" AI platform for broader application in manufacturing.
📊 競品分析▸ Show
| Feature/Company | Sumitomo Rubber/Fujitsu (AI Surrogate Model) | NEXEN TIRE (AI Performance Prediction System) | Dassault Systèmes SIMULIA (AI for Tire Optimization) | Altair (Physics AI) | NVIDIA PhysicsNeMo / Siml.ai | Energent.ai |
|---|---|---|---|---|---|---|
| Focus | Tire structural analysis (deformation, contact characteristics) | Key tire performance metrics (fuel efficiency, noise, handling, grip) | Tire design optimization (new sizes, trade-offs) | Durability and stiffness of automotive components | General physics-based AI surrogate modeling | AI-accelerated FEA, unstructured data processing |
| Technology | Graph Neural Network (GNN) surrogate model | Machine Learning | AI, Reduced-Order Models (ROMs) | Machine Learning | Physics-informed ML, Deep Learning | AI-powered platforms |
| Speedup | 90% reduction (45 min to 5 min) | "Quickly and accurately forecast" | "Speeding up design iterations" | 30% reduction in design/solution times | Near-real-time latency | Up to 100x faster structural predictions |
| Accuracy | 87.7% average accuracy for tire-road contact shape | Not specified (high accuracy claimed) | Not specified (precise answers claimed) | Accurate predictions | High-fidelity | Not specified for simulation, 94.4% for data parsing |
| Hardware | Optimized for Fujitsu-Monaka CPU | Not specified | Not specified | Not specified | Optimized for NVIDIA GPUs | Not specified |
| Availability | Practical implementation at Sumitomo Rubber by April 2027; Fujitsu Kozuchi platform for others | Established in 2022 | Commercial solutions available | Commercial solutions available | Open-source platform (PhysicsNeMo), commercial (Siml.ai) | Commercial platform |
🛠️ 技術深入
- The AI surrogate model is built upon a Graph Neural Network (GNN) algorithm.
- It is trained using accumulated finite element method (FEM) analysis results and Sumitomo Rubber's tire design expertise and actual design data.
- The model's function is to rapidly predict solutions to the governing equations used in FEM analysis.
- The proof-of-concept specifically evaluated tire deformation behavior and contact characteristics, including contact shape and pressure distribution under road contact conditions.
- The technology is optimized for Fujitsu's next-generation Arm-based CPU, FUJITSU-MONAKA, which is designed for high performance and energy efficiency.
- Fujitsu's broader AI Solver platform, which may underpin this solution, aims to convert physics-based simulators into AI simulators, achieving speedups from hours to milliseconds with minimal discrepancy.
- The solution is intended to be integrated into a tire design support tool, making it accessible to designers without requiring specialized machine learning expertise.
🔮 前景展望基於引用來源的 AI 分析
The widespread adoption of AI surrogate models will significantly shorten product development cycles across the automotive industry.
By reducing simulation times from hours to minutes, engineers can conduct far more design iterations, leading to faster innovation and time-to-market for new vehicles and components.
Fujitsu's "Fujitsu Kozuchi" platform will become a leading solution for AI-accelerated CAE in manufacturing.
Fujitsu's plan to offer this GNN-based AI inference platform, optimized for FUJITSU-MONAKA, to other manufacturing industries positions it as a versatile tool for broader industrial application.
Tire design will become more democratized within manufacturing companies.
The planned design support tool will allow designers to directly use the AI-accelerated analysis without specialized expertise, broadening access to advanced simulation capabilities.
⏳ 時間線
2022
Sumitomo Rubber Industries and NEC began co-creation activities to develop an AI platform for enhancing tire development capabilities.
2023-04-20
Fujitsu launched its "Fujitsu Kozuchi" AI Platform, providing access to various AI and ML technologies.
2024-12-11
Sumitomo Rubber Industries engaged Rockwell Automation to further its global digital transformation strategy for manufacturing process optimization.
2025-11-26
Sumitomo Rubber Industries and NEC announced results of demonstrations utilizing pseudo-quantum annealing and NEC Material Discovery Solution for new material discovery.
2026-06-03
Sumitomo Rubber Industries and Fujitsu announced the joint development of an AI surrogate model that reduces tire structural analysis time by approximately 90%.
📎 來源 (20)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
📰
AI 週報
閱讀本週精選 AI 大事摘要 →
👉相關動態
AI 策展新聞聚合。所有內容版權歸原始發布者所有。
原始來源： ITmedia AI+ (日本) ↗
每週電子報
每週一封，可隨時退訂。