Investigation of solutions of sample problems in the automotive industry using artificial intelligence methods

Many different defect modes are encountered in the current processes in the automotive manufacturing sector. Temporary and permanent measures are tried to be taken to eliminate these defects. Most of the time, due to the nature of the processes, zero defect conditions cannot be created in the activities performed. In order to eliminate the weakness here, measures are taken to prevent the error from being sent to the next process or to the internal and external customers with 100% visual control at the human initiative. However, due to the defects that cannot be detected in the field defect scans, extremely high costs of poor quality are encountered, depending on the severity. In the modern approach, defect detection with high accuracy rates can be achieved by using different methods of machine learning instead of visual controls in defect detection. For example, due to its superior image processing, feature extraction capability, convolutional neural network (CNN) has been widely researched and applied in the field of intelligent defect diagnosis and has gained superior performance compared to other traditional machine learning methods. In addition, in the determination of process parameters, the most ideal parameter results can be obtained by using the artificial neural networks (ANN) method by making use of the available field data. In this study, an innovative approach has been developed that allows defect detection in manufacturing processes. In sample studies, In the determination of spot welding parameters, nugget diameter estimation was made with existing parameters using artificial neural networks (ANN). Here, comparative results are drawn according to the different regression analysis results. In addition, high accuracy rates have been achieved by using convolutional neural networks (CNN) in machine learning to detect spot and arc metal arc welding defects, sheet metal crack defects and vehicle visual defects.