Soft measurement modeling method of semi-supervised dynamic soft measurement network

Abstract

The invention discloses a soft measurement modeling method of a semi-supervised dynamic soft measurement network. The soft measurement modeling method is specifically implemented according to the following steps: performing denoising and redundancy elimination processing on training set data based on CEEMD and Isomap methods; serializing and normalizing the training set data; and establishing a soft measurement model of the semi-supervised dynamic soft measurement network based on the training set data. According to the modeling method disclosed by the invention, the noise and redundancy of the data are removed by using a CEEMD and Isomap combined method; CEEMD has completeness, and the result of CEEMD has no obvious modal aliasing phenomenon; the Isomap has a very strong nonlinear featuretransformation capability, and the advantages of the two methods are integrated, so that noise and redundancy in original data are effectively removed, information loss is reduced to the maximum extent, serialization processing is carried out on the data, historical data is introduced for dynamic modeling, and compared with a traditional soft measurement method, the method has the advantage thatprediction of variables by using the model is more accurate.

Description

technical field [0001] The invention belongs to the technical fields of intelligent signal processing and industrial artificial intelligence, and in particular relates to a soft sensor modeling method of a semi-supervised dynamic soft sensor network. Background technique [0002] In complex industrial processes, due to the harsh production environment, complex working conditions, limited detection technology or cost, the key quality variables in the process often cannot be measured reliably and in real time. In order to overcome this problem, soft-sensing technology emerged as the times require. It takes auxiliary variables that are easy to measure in the process as input, and the leading variable that you want to measure as output. an accurate estimate of . [0003] With the development of information technology and the widespread application of distributed control systems in complex industrial processes, massive industrial process big data can be collected and used to est...

AI Technical Summary

Problems solved by technology

[0003] With the development of information technology and the widespread application of distributed control systems in complex industrial processes, massive industrial process big data can be collected and used to establish soft sensor models, but external environmental disturbance factors and fluctuations in the process itself include changes in raw material composition, Factors such as random interference in the process of data transmission and storage lead to industrial data often containing a large amount of data noise. At the same time, massive industrial production data inevitably introduces data redundancy problems, that is, the same production conditions repeatedly appear in the production process and The collinearity between different auxiliary variables, the noise and redundancy contained in the data will seriously affect the accuracy of the data-driven soft sensor model, and it needs to be removed in the preprocessing stage

In addition, industrial process data is often a continuous time series. If static soft sensor modeling is performed on it, the context of the data cannot be captured, which often leads to low accuracy and poor robustness of model estimation in practical applications.

Images