A SVM-based Quality Assessment using Thermal Image DATA in Sealer Dispensing Process

Abstract

As manufacturing environments are getting global and complex, the automated and systematic ways of product quality monitoring are becoming more popular in the global manufacturing sector, especially in automotive industry. In a number of manufacturing processes, however, the process of quality assessments in production is still highly dependent on manual inspections due to the complex patterns of defects and lack of input data in both products and processes. In this paper, thus, a case of the automated quality assessment system in an automotive company is investigated. To enhance the quality monitoring capability of the production system, the infrared thermal image, which includes nominal dimensions and temperatures, is analyzed using support vector machine (SVM) algorithm. The thermal image data can provide multi-dimensional quality information which may not be possible to be assessed by human visions. At the end of this paper, a sealer dispensing in an automotive parts assembly process is illustrated to verify the applicability of the system.