Soft-sensor Technology And The Application In The Crude Oil Dehydration Process

This paper analyzes the soft-sensing technique of modeling methods,variable selection and data treatment, etc. According to various industry processes, have different modeling methods and relevant basic theory, and make a brief description about advantages and disadvantages. In the paper, we have a brief depiction about the crude oil dehydration process, and the significance and importance of setting up the soft-sensor model of oil-water interface highness.By studying a variety of soft-sensor modeling methods, here selected two modeling methods which based on artificial neural network theory and the support vector machine(SVM) which based on statistical learning theory. For the problem in the dehydration of crude oil production process, here starts research and application. Here founds the oil-water interface highness model, which using the selected methods.Based on the characteristics and improved algorithm of the SVM, this paper brings forward a method—hybrid kernel function of the weighted least squares support vector machine soft-sensor model. The model achieves the better effect than above modeling methods.