Elimination of scrap (defected product) is the key objective of every process engineer involved designing manufacturing workflow. Multiple stations with a variety of non-destructive testing methods are built on assembly lines to detect scrap early on. Quality control is therefore important not only to ensure that defective pieces don’t get to the end customer but also to make efficient use of resources. Major savings could be achieved if some of the early tests proved to be sufficient in detecting scrap products as the tests further down the line could be removed altogether.
Our customer, a leading provider of lightweighting solutions for the global automotive industry, understands the need for a data-driven approach to this challenge. One of the company’s plants, well known in the group with its track record of innovations, chose Cognexa as a partner in Big Data and Machine learning domain.
The aim of the pilot project was to enhance the efficiency of the production process of mold products by reducing scrap percentage on one of its production lines. The rapid experimenting capability of the customer’s process engineering team was boosted by the insights extracted from the production data by Cognexa data science team.
Our team analysed impact of different production parameters on the quality and combined the findings with the results obtained from internal and external testing of products. Using machine learning-based predictive modelling and root cause analytics, we have created a model to determine the percentage of scrap in a product batch considering given values of production parameters. We also modelled the results of external tests and the relation between their results and assessed hypotheses of the process engineers.
The cooperation between Cognexa and the customer is still ongoing with a focus on different projects now. The success of the project also sparked interest in the customer’s international group and future projects are now being prepared.
Automotive Components Manufacturer
Root cause analysis and modelling of quality testing results
A major production process change is being discussed at a group level thanks to our data insights