Support System of Acceptance-Rejection Decision for Incoming Quality Control Process

Abstract

In this study, predictive support systems have been designed for a structure in which takes into account the expert opinions of employees for decision making in addition to four input factors in a yarn quality acceptance process. The first technique is a hybrid genetic algorithm which was previously designed and validated for the classification study and adapted to the current problem. The other method is a hybrid radial based function (RBF) based on neural networks which is adapted to the problem and coded. During the process of classifying the actual production data for the accept-reject decision, accuracy rate of more than 90% has been achieved with two techniques developed and also some other methods which are well-known in literature were used for performance comparison. After the verification of the hybrid genetic algorithm performance, the best chromosome that was obtained, used as the model for classification estimation. According to the proposed methodology, the selected attribute values were multiplied by the determined coefficients and compared with a threshold value, and an acceptance-rejection decision could be made with a reasonable accuracy rate. The contribution of the article to the literature can be evaluated in two ways. The first one is comparing the classification performance of the proposed hybrid genetic algorithm with the hybrid neural networks method, and the second one is utilizing the best chromosome of the proposed genetic algorithm as a support system for yarn quality acceptance process.

Keywords

• Hybrid genetic algorithm
• Hybrid RBF Classification