Quality Mining in a Continuous Production Line based on an Improved Genetic Algorithm Fuzzy Support Vector Machine (GAFSVM)

Abstract

Nowadays, instant decision making through applying available vast raw data has been considered as one of the main challenges in various industries. Accurate and quick decisions can lead to enhancement of the product quality, especially in continuous production lines. Meanwhile, the quality outputs take several hours to be determined. In addition, stoppages during production procedure could result in changing the feature settings and affecting the product quality. Herein, a new hybrid approach has been proposed to address the mentioned challenges faced by industries. In the first step of the suggested model, the data mining approach was utilized to predict the possible required stoppages and their duration. Then, the real production time of order was obtained through the improved time series classification model based on fuzzy support vector machine and genetic algorithm. Afterward, the product quality level was predicted considering the start and end times of the considered order. The efficiency and accuracy of the presented model were confirmed through implementation on an iron-making unit of a steel company. According to the obtained data, the mentioned model could be beneficial to approach remarkable performance and improved product quality management in various industries.