Quality and Productivity Enhancement through AI Analysis of Process Defect Data

This is Impix's AI solution that predicts the quality of injection molding processes and improves productivity.



1. Pain Point 

Lack of objectivity in AI solutions

- Although work standards for injection molding processes have been defined, the process of deriving these standards is not objective and lacks specificity.

- When issues arise, adjustments to equipment settings are made based on workers' expertise without objective guidelines, and defect causes are identified through subjective judgment.



Limitations in the process

- Insufficient data analysis capabilities make it difficult to resolve chronic quality defects in injection molding equipment.

- Inadequate digitization of workers' experience and expertise limits process standardization.

- Inefficient problem solving due to inadequate data-based analysis and decision-making systems.

- Limited ability to improve work analysis and problem-solving capabilities due to insufficient data collection and analysis systems.

2. Construction Details

Smart Factory Construction

- Conducted exploratory data analysis (EDA) on existing data to perform data exploration and correlation analysis.

(Analyze the relationship between manufacturing environment variables such as speed, temperature, and time in the automatic injection molding process and production efficiency)

- Build a dataset through data processing such as removing missing values.

- Derive a process optimization model by applying appropriate algorithms and models.

- Support easy application in the field by utilizing the visualization and simulation functions of the solution.

 

3. Effects of Smart Factory Construction

Increased Production Volume

- Monthly production volume increased by 9.4%

- Before implementation: 188,747 units → After implementation: 206,190 units

Improved Prediction Accuracy (F-1 Score)

- 0.8 (80% improvement in accuracy)