Application of pattern recognition to ethylene production optimization

Abstract In this paper, a pattern recognition method has been applied to solving the problem of the optimization of ethylene production process. The basic point of the presented method is that the technological parameters are used as feature variables to construct the pattern space, and all samples are divided into two classes according to the production target to be optimized. In order to find out the key factors influencing the target, the method of feature extraction is adopted to reduce the dimensionality of the pattern space of the technological parameters. By using the Fisher rule and the fractional correction rule, a recognition model has been found, which can distinguish the operating conditions of high-quality product from that of low-quality product. Based on the recognition model, some specific methods are suggested for optimal production.