Abstract
The manufacturing process of car body parts needs to be adaptable during production because of fluctuating variables; finding the most suitable settings is often expensive. The cause-effect relation between variables and process results is currently unknown; thus, any measure taken to adjust the process is necessarily subjective and dependent on operator experience. To investigate the correlations involved, a data mining system that can detect influences and determine the quality of resulting parts is integrated into the series process. The collected data is used to analyze causes, predict defects, and optimize the overall process. In this paper, a data-driven method is proposed for the inline optimization of the manufacturing process of car body parts. The calculation of suitable settings to produce good parts is based on measurements of influencing variables, such as the characteristics of blanks. First, the available data are presented, and in the event of quality issues, current procedures are investigated. Thereafter, data mining techniques are applied to identify models that link occurring fluctuations and appropriate measures to adapt the process so that it addresses such fluctuations. Consequently, a method is derived for providing objective information on appropriate process parameters.
Highlights
The car body part production process is affected by several variables [1] that cause quality fluctuations
Using the results of 3.2 a method is proposed that efficiently determines the necessary adjustments of one control variable
Experimental results For validation, the proposed method is applied to the series process data from a data set involving the production of 52,753 bottom plates of a BMW 3 series sedan
Summary
The car body part production process is affected by several variables [1] that cause quality fluctuations. When a quality defect appears, the process needs to be manually adjusted using control variables until parts of acceptable quality are produced. Digitization in the context of the so-called “industry 4.0” offers the potential to solve these problems. One aspect of this digitization involves the automatic exchange of information between machines and products, which creates production processes that continuously produce large amounts of data [2]. The analysis of this data enables the understanding and the modeling of complex relations.
Sign-in/Register to view highlights & summary 
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call