AI Case Studies
R&D
AutoML helps R&D analyze data and utilize predict models, accelerating new product development cycles and enhancing innovation efficiency. By analyzing historical product quality data, it continuously optimizes parameter models, aiding R&D teams in formulating more precise strategies.
Application Scenarios
Formula Optimization, Parameter Recommendation
Optimizing R&D Formulas
The R&D department often conducts numerous experiments during the development of new products and the improvement of existing ones, involving various complex and variable factors.
Pain Points Analysis:
- High Experiment Costs: Traditional experimental design methods require significant time, labor, and resources.
- Numerous Variables: It’s challenging to accurately identify and optimize the numerous variables and their interactions using traditional methods.
- Reliance on Experience: Formula design relies on the experience of senior engineers, which is difficult to pass on to new employees.
Outcome Benefits:
- Feature Importance Analysis: Quickly identify key feature parameters affecting R&D goals, helping researchers focus their efforts.
- Simulation Function: Analyze the impact of multiple variables on targets, shortening the R&D time.
Parameter Recommendation During NPI Stage
When transitioning from R&D to mass production, production parameters need adjustment to ensure smooth entry into mass production while maintaining high quality and productivity.
Pain Points Analysis:
- Lack of Data: Historical data for new products in this stage is limited, making effective parameter setting and optimization challenging.
- Production Stability Challenges: The production process is not yet fully stable, requiring repeated adjustments and optimizations, increasing trial and error costs.
- Cross-departmental Collaboration Difficulties: The NPI stage requires collaboration between R&D, manufacturing, and quality departments. Relying solely on personal experience can hinder parameter optimization due to communication and collaboration inefficiencies.
Outcome Benefits:
- Learning from Historical Data: Combine historical parameter adjustment records of similar products to learn and focus on parameters affecting product quality, reducing trial and error time.
- Visualization of Optimization Results: Promote effective communication and collaboration between R&D, manufacturing, and quality departments.
First-piece Inspection Hit Rate Prediction
In the mass production stage of batch production types, small batch inspections are usually conducted to ensure product quality and verify the correctness and stability of production parameters.
Pain Points Analysis:
- Unstable Incoming Material Quality: Quality fluctuations in different batches of incoming materials affect the first-piece inspection, requiring parameter adjustments for incoming materials.
- Lack of Data Support: Traditional methods rely on experience for parameter settings, which cannot effectively record and transfer judgment.
Outcome Benefits:
- Correlation Analysis: Identify the correlation between incoming materials and parameter control from historical production data, improving the first-piece hit rate by 7%, reducing trial and error times and adjustment time.
- Continuous Optimization: Continuously collect and analyze incoming and production data to constantly optimize the parameter recommendation model, achieving continuous improvement of the production process.