Distributed industrial process monitoring method based variable weighting PCA (Principal Component Analysis) model

Abstract

The invention discloses a distributed industrial process monitoring method based on a variable weighting PCA (Principal Component Analysis) model. The method mainly solves a problem that hidden useful information cannot be comprehensively explored when a PCA model is built because a traditional PCA based fault detection method fails to sufficiently consider the correlation difference between measurement variables. According to the method, diversified weighting is performed on training data by using the correlation between each measurement variable and other measurement variables in the process, and a corresponding PCA fault detection model is built for the weighted data so as to carry out distributed process monitoring on a production object. The correlation strength between the measurement variables can be reflected in a prominent manner by implementing weighting on the training data by using the size of a correlation value. On this basis, hidden information extracted by using a PCA algorithm can better describe the operating state of the corresponding measurement variable. In addition, a plurality of PCA fault detection models are utilized to implement distributed process monitoring, so that the diversity of the extracted process data hidden information can be well ensured, and thus comprehensive description for the process operating state is realized. Compared with a traditional PCA based fault detection method, the method disclosed by the invention has a more comprehensive process operating state description ability, and can acquire a more reliable and more excellent fault detection effect.

Description

technical field [0001] The invention relates to an industrial process monitoring method, in particular to a distributed industrial process monitoring method based on variable weighted PCA model. Background technique [0002] In the entire integrated automation system, the fault detection system is an indispensable part, because real-time monitoring of the operation status of the production process is a necessary means to ensure production safety and maintain stable product quality. With the rapid development of science and technology, modern industrial processes are usually composed of multiple interleaved production units, and physical models that can accurately describe their operating mechanisms cannot be obtained. However, industrial processes can measure and store vast amounts of data samples thanks to the widespread use of various advanced instrumentation and computing technologies. Therefore, implementing fault detection throughout an integrated automation system no ...

AI Technical Summary

Problems solved by technology

This data preprocessing method has a defect for analysis algorithms such as PCA that aim to mine the correlation of data variables: the correlation between data variables is different, if the correlation difference between data variables is not effectively analyzed On balance, the latent information extracted by the PCA algorithm cannot fully and effectively describe the state of the process operation

Therefore, the fault detection effect of the process monitoring method based on PCA, especially based on distributed PCA, still needs to be further discussed.

Images