AI-based molding process line optimization model derivation demonstration

Industrial characteristics

- Analysis and standardization of the causal relationship between environmental factors such as pressure, temperature, and humidity, which vary depending on the characteristics of raw materials, and production efficiency are required.

- Issues exist regarding yield reduction due to quality variations and high defect rates despite high operating rates and production volumes in high-pressure molding processes.

Data and System Aspects

- Improvement is needed to reduce reliance on the experience and subjective judgments of highly skilled workers after MES-based data collection.

- Criteria for identifying causes and adjusting equipment settings are absent when defects occur.

Equipment and infrastructure aspects

- There is a lack of standard work procedures and equipment setting guidelines for the high-pressure molding process. 

- There is a need for correlation analysis between defect types and key variables.  

- There is a need to establish a data-based process optimization and rapid feedback system.

2. AI Solution

Impix's A2LAB AI Solution

- Improved learning performance of defective data through EDA (exploratory data analysis), missing value removal, and SMOTE-based oversampling.

- Real-time data linkage and AI analysis automation using MES and sensor data.

- AI-based process simulation, visualization, and provision of optimal equipment setting values to create a structure that is easy for end users to utilize.

3. Construction Goals

Detailed Goals

- Promote optimization and standardization of high-pressure molding processes through AI-based data analysis.

- Establish a data-based decision-making system to identify the causes of defects and lay the foundation for improvement.

- Lay the foundation for a competitive intelligent smart factory through quality and yield improvements.

- Secure the basis for verification of company-wide expansion and applicability to other processes and factories.

4. Construction Details

Data Integration Management

- Build and manage a dataset for AI analysis by integrating MES, sensor, and manual data.

AI-based analysis and modeling

- Analyze the correlation between variables and quality and derive key variables.

- Build and validate a process optimization model based on XGBoost.

Simulation and field application

- Build analysis and simulation functions based on VM servers and BI tools.

- Utilize quality and yield prediction functions prior to adjusting equipment settings through process simulation functions.

 

5. Construction Effects

Monthly Production Volume Increase

 5,087 EA → 6,899 EA

Process Defect Rate Reduction

 33.0% → 14.0%