Implementing data analytics for improved quality in manufacturing: a case study
Examensarbete för masterexamen
Production engineering (MPPEN), MSc
The possibilities for extracting important information from big data have accelerated over the last few years. More manufacturing companies are realizing the potential of advanced data analytics, such as machine learning, and how this can be used to improve productivity and sustainability for future competitiveness. It is how ever often not clear how to approach the adoption of data analytics and overcome the necessary challenges before real value can be created using these techniques. This thesis proposes what some of the most relevant challenge areas are to improve the conditions of conducting successful pilot projects for data analytics in manu facturing. A suitable methodology for the implementation of data analytics is also presented, which is evaluated in a case study at a machining department at Volvo. The problem being solved in this case study was related to product quality improve ment, where to goal was to derive which influencing factors in a machining process have the highest importance for the quality outcome. A state-of-the-art literature review was performed to evaluate which requirements are necessary efficiently uti lize data analytics and which methodologies can be used for the performing use cases. An adapted version of an implementation methodology, the Cross Industry Standard Process for Data Mining (CRISP-DM), was developed to be better suited for the specific challenges in manufacturing. This methodology was thereafter used and evaluated in the case study. A combination of interpretable machine learning models was used in conjunction with association rules to describe which influenc ing factors are the most relevant for the quality outcome. From the results, it was possible to identify improvement areas that Volvo needs to address for better use of data analytics. Important findings were that the internal knowledge needs to be increased and that certain technical challenges need to be further developed, such as connectivity, system integration, and traceability. These results can help Volvo and other manufacturing companies in a similar development stage to understand how to prioritize efforts for succeeding with the implementation of data analytics. It especially gives insights into how to deal with use cases related to quality improve ment and how to increase the interpretability during the implementation of data analytics.
Data analytics, machine learning, manufacturing, quality improvement.