Gaussian process regression soft measurement modeling method based on EGMM (Error Gaussian Mixture Model)

Abstract

The invention discloses a gaussian process regression soft measurement modeling method based on an EGMM (Error Gaussian Mixture Model), which is used for a complex and changeable chemical process with non-gaussian noise. Prediction errors are frequently generated by a soft measurement prediction model established in an industrial process, however, the model prediction errors frequently contain rich useful information, and therefore, information can be extracted from the prediction errors so as to compensate the output of the model, thereby improving the established soft measurement model. Firstly, appropriate variables are selected to form error data, so as to be optimized to obtain appropriate number of gaussian components; then fitting is performed on the error data by using the EGMM; when new data arrive, prediction output is performed by using established GPR (Gaussian Process Regression), the mean conditional error is obtained through the EGMM, and the output is compensated, so as to obtain more accurate results. Key variables can be accurately predicted, thereby increasing the quality of products and reducing the production cost.

Description

technical field [0001] The invention relates to a Gaussian process regression soft sensor modeling method based on EGMM, and belongs to the field of complex industrial process modeling and soft sensor. Background technique [0002] In modern industrial processes, data-driven soft-sensing modeling methods have received more and more attention. Some commonly used soft sensor modeling methods such as partial least squares (PLS) and principal component analysis (PCA) can well deal with the linear relationship between input variables and output variables. Artificial neural networks (ANN), support vector machine (SVM), and least (support vector squares support vector machine, LS-SVM) can effectively deal with the nonlinear relationship of the process. [0003] In recent years, Gaussian process regression (GPR), as a non-parametric probability model, can not only give the predicted value, but also obtain the trust value of the predicted value to the model. The present invention s...

AI Technical Summary

Problems solved by technology

Although these modeling methods can effectively deal with the high nonlinearity of the process and the high-dimensional mapping relationship between input and output, these data models are often established by assuming that the modeling error conforms to the Gaussian distribution.

In fact, industrial processes often contain different stochastic distributions, diverse measured scatter and non-measured inputs (hidden inputs), and the modeling is often very complex

Images