Just-in-time learning soft measurement modeling method based on Bayes Gaussian mixture model

Abstract

The invention discloses a just-in-time learning soft measurement modeling method based on a Bayes Gaussian mixture model, and belongs to the field of complex industrial process modeling and soft measurement. The method is used for a time varying industrial process with nonlinearity and non-Gaussianity; through a strategy of updating localities in real time online, a Bayes information criterion isadopted to determine an optimal Gaussian ingredient number; when new test data comes, a posterior probability that the new test data belongs to each Gaussian ingredient is calcuated, a Mahalanobis distance between the new test data and training data is solved, and the posterior probability and the Mahalanobis distance are blended to serve as a similarity index; and finally, one group of data withthe highest similarity is selected from the original training sample to establish a current GPR (Gaussian Process Regression) model, and model output prediction is carried out to achieve an effect onimproving product quality and lowering production cost.

Description

technical field [0001] The invention relates to an instant learning soft sensor modeling method based on a Bayesian Gaussian mixture model, which belongs to the fields of complex industrial process modeling and soft sensor. Background technique [0002] For some industrial processes that are nonlinear, time-varying and non-Gaussian, and the requirements for product quality in the process are constantly increasing, it is necessary to strictly monitor and control some process variables that directly determine product quality. However, due to the high price of some measuring instruments or the restriction of technical conditions, these variables cannot be measured by online instruments. [0003] In order to solve these problems, it is possible to estimate and predict by establishing a soft sensor model. Commonly used soft sensor methods include partial least squares (PLS), artificial neural networks (ANN), support vector machine ( support vector machine, SVM), etc. PLS can de...

AI Technical Summary

Problems solved by technology

[0006] However, for some non-Gaussian time-varying industrial processes, the traditional JITL method selects similar data based on the Euclidean distance or the similarity criterion combined with the angle, which cannot fully consider the non-Gaussian nature of the process data

Images