Page 519 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

• Solution: GAC group has built a "1+3+3+N" unified architecture system, that is, one
digital foundation, three intelligent platforms (IoT cloud-edge collaboration platform,
industrial big data platform, and industrial AI large model platform), three scenarios
(intelligent R&D, intelligent manufacturing, and intelligent supply chain) and N AI-based
applications for automobile manufacturing. The scenarios are described in detail below. 4.5: Manufacturing

1. Intelligent generation of automobile styling concept design: Using AGI’s text-to-image/
image-to-image technology to generate automobile styling concept design, , aiming
to address the issues of long concept design cycles and insufficient design creativity in
traditional design processes, thereby reducing product homogenization and shortening
project development cycles.
2. Intelligent analysis of vehicle design failure modes: Using “LLM + GraphRAG” to
understand the vehicle structure and function requirements, automatically generate
failure modes and preventive measures, and improve the efficiency of vehicle design
failure analysis. RAG includes vehicle body component information, function and failure
mode correlation data, internally accumulated design guidelines, historical quality issues,
research papers, technical specifications, design validation plans, etc. A function failure
analysis, measures, and follow-up implementation that originally required 6 person-days
can be shortened to 2 person-days.
3. Intelligent visual quality inspection: Using CV large model to realize the appearance
defect inspection of stamped sheet metal parts, paint surface inspection, final assembly
specification inspection, etc., to reduce low-value and repetitive manual operations,
reduce labor risk and intensity of employees.
4. Predictive maintenance of devices: Using a predictive large model and LLM, 24-
hour monitoring and analysis of equipment health status can be achieved. Building
a maintenance knowledge base to shorten personnel training cycles and improve
equipment availability.
5. Intelligent production scheduling: Using solvers and multi-objective optimization heuristic
algorithms are used to realize automatic production scheduling, reduce inventory, and
respond to unexpected events. The production capacity will be improved by 10% and
the production scheduling time will be shortened by 60%.

Use Case Status: The use case is part of a larger research project�

Partners

Huawei: Provides unified digital infrastructure, R&D tool chains, basic large model, IoT cloud-
edge collaboration platform, industrial big data platform, and industrial AI large model platform.
GAC Group: Focus on the development of AI application scenarios in the "R&D, production,
supply, sales, and service" field, and the establishment of high-quality dataset.

483