Predicting Enamel Layer Defects in an Automotive Paint Shop

Abstract

The appearance of the painted surface of the vehicle is key in the quality that the automotive customer perceives. The assurance of this quality starts in the automotive paint shop and compromises the effectiveness of the painting process as every paint defect is reworked. This entails material and labour costs, reducing the efficiency of the process and affecting the competitiveness of the product. To improve the efficiency while guaranteeing the quality, predictive control rather than corrective must be implemented. In order to achieve this control, a predictive model of quality is needed. As a first step to generate said model, this article demonstrates the correlation between the variables of the enamel coating process and the quality of the paint film of the vehicle. As there are no available application examples in the industry, a procedure is proposed in which the necessary steps for the creation of an industrial data set and a predictive model of quality are defined. The procedure is tested in an automotive paint shop. As a result, relevant variables for the quality assurance are identified and the correlation between process variables and the resulting quality is verified, concluding that the implementation of predictive control in the process is feasible.

Keywords

• Predictive modelling
• automotive industry
• manufacturing digitization
• industrial data set