Soft measuring meter moduling method based on supporting vector machine

Abstract

The method includes two parts: projective relation between input and output in modeling of soft measuring instrument of supporting vector machines is completed by vector machines; controlled inputs of measurable variables and objects as well as measurable output variables of objects are as input variables for soft measuring instrument, and optimal evaluation of variables is as output. Selected from input variables, a group of second order variables closely related to leading variables is as input of supporting vector machines. Off-line anacom values or measured values in large sampling interval are as output of soft measuring model. First, using posterior distribution and extremum principle of normalized parameters through iteration determines normalized parameters of standard support vector machine and least square support vector machine. Then, under Bayesian third criterion, iteration determines kernel parameters for the said vector machines.

Description

technical field [0001] The invention relates to a modeling method of a soft measuring instrument, in particular to a modeling method of a soft measuring instrument based on a support vector machine. For use in the field of measurement technology. Background technique [0002] In the modern process industry, a large number of key parameters such as process status and product quality lack online direct measurement means. This has become a bottleneck restricting the further improvement of production safety, product quality, quality and production efficiency. Soft sensing technology is an effective way to solve such problems. [0003] Soft sensor technology is a research hotspot in the field of control. Its core technology is to establish the model of soft measuring instruments. At present, the modeling methods of soft measuring instruments mainly include mechanism modeling, multivariate statistical methods, Kalman filtering methods, artificial neural neural networks, model-b...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a kind of soft sensor instrument modeling method based on support vector machine for the above deficiencies and defects that exist in the existing soft sensor modeling technology, provide the selection method of the optimal model of support vector machine, support vector Machine (including standard support vector machine and least squares support vector machine) optimal model determination method, so that it overcomes the problem of model determination difficulties in the application of support vector machine, and establishes for the soft sensor modeling of support vector machine a solid foundation and

Images