[3M] Machine Learning - Injection Moulding Machine

Last updated on 

Worked in a team of 6 to develop a machine learning algorithm to analyze material properties of injection molded plastic parts to quickly (130 parts) predict appropriate process parameters, without the need for a skilled operator or trial and error.

Python, Scikit-learn, Seaborn

  1. Regression models to predict the ultimate tensile strength of a part

2.  Classification models to predict if a part was good or bad

3.  Machine vision to detect bad parts on assembly line

The Process

Saxena, S. (2018, December 10). EMBEDDING CONFORMAL COOLING CHANNELS IN INJECTION MOULDING USING METAL ADDITIVE MANUFACTURING. ADDITIVE MANUFACTURING/ 3D-PRINTING. https://additivemanufacturingindia.blogspot.com/2018/12/

Process parameters

Material properties

(right) Tensile test used to determine material properties
Visual tests

Cleaning data

Checking for outliers

Regression

Classification

k-fold cross validation

Machine vision

Transfer Learning using Xception, ResNet50. Identification using YOLO