Non-gaussian process monitoring method based on novel dynamic independent component analysis

Abstract

The invention discloses a non-gaussian process monitoring method based on novel dynamic independent component analysis, and aims to combine the advantages of a dynamic internal principal component analysis model which can deal with autocorrelation dynamic data with an independent component analysis model which can deal with non-gaussian data. Specifically, the non-gaussian process monitoring method includes the steps that firstly, a dynamic internal principal component analysis algorithm is used for correspondingly extracting autocorrelation dynamic characteristic components and cross-correlated static characteristic components; secondly, after whitening processing of the characteristic components, combined whitening characteristic components are used as initial independent components anda dynamic independent component variable model is obtained iteratively; and finally, based on the dynamic independent component variable model, dynamic non-gaussian process monitoring is implemented.In conclusion, according to the non-gaussian process monitoring method based on novel dynamic independent component analysis, the ability of the dynamic internal principal component analysis algorithmseparately extracting dynamic components and static components is utilized, and an independent component analysis algorithm capable of extracting non-gaussian characteristic components is further combined. Therefore, the non-gaussian process monitoring method based on novel dynamic independent component analysis is the feasible dynamic non-gaussian process monitoring method.

Description

technical field [0001] The invention relates to a data-driven process monitoring method, in particular to a non-Gaussian process monitoring method based on novel dynamic independent component analysis. Background technique [0002] With the rapid development of advanced measurement technology and computing technology, modern industrial processes are gradually moving towards digital management. It is against this background that the industrial "big data" boom has emerged. Since production process objects can store and measure massive amounts of data offline and online, and these data contain information that can reflect the operating status of the production process, the use of sampled data to monitor the operating status of the process has been favored by many scholars. In fact, both academia and industry have invested a lot of manpower and material resources in the research of process monitoring methods with fault detection and diagnosis as the core task. In the field of d...

AI Technical Summary

Problems solved by technology

Although the DiPCA algorithm can mine autocorrelation features, it fails to retain autocorrelation features, and at the same time fails to further mine autocorrelation non-Gaussian independent component information.

Therefore, the non-Gaussian process monitoring performance of the DiPCA algorithm needs to be improved

Images