• SYSTEM ENGINEERING • 上一篇 下一篇
张英苏宏业刘瑞兰褚健
ZHANGYing;SUHongye;LIURuilan; CHUJia
摘要: In the past few years, support vector machines (SVMs) have been applied to many fields, such as pattern recognition and data mining, etc. However there still exist some problems to be solved. One of them is that the SVM is very sensitive to outliers or noises because of over-fitting problem. In this paper, a fuzzy support vector regression (FSVR) method is presented to deal with this problem. Strategies based on k nearest neighbor (kNN) and support vector data description (SVDD) are adopted to set the fuzzy membership values of data points in FSVR. The proposed FSVR soft sensor models based on kNN and SVDD are employed to predict the concentration of 4-carboxy-benzaldehyde (4-CBA) in purified terephthalic acid (PTA) oxidation process. Simulation results indicate that the proposed method indeed reduces the effect of outliers and yields higher accuracy.