Convolutional self-encoding fault monitoring method based on batch imaging

Abstract

The invention discloses a convolution self-encoding fault monitoring method based on batch imaging, and belongs to the technical field of batch process fault monitoring. The method comprises two stepsof off-line modeling and on-line monitoring. The offline modeling step comprises the following steps: firstly, normalizing three-dimensional data of the intermittent process; then, taking the two-dimensional array of each batch as an image to be directly input into a convolutional auto-encoder (CAE) to carry out deep unsupervised feature learning; and finally, constructing statistics and corresponding control limits for the features learned by the CAE by utilizing a support vector machine. The online monitoring step includes: normalizing the collected data, and carrying out batch filling; inputting the normalized and filled batch graph into the trained CAE to learn features; and calculating an online statistic, and comparing the online statistic with an offline control limit. Compared with the prior art, the technical scheme provided by the invention avoids information loss caused by data expansion, does not need to divide stages to reduce modeling workload, deeply extracts change characteristics of process variables, and reduces false alarm and missing report rate of intermittent process monitoring.

Description

technical field [0001] The invention belongs to the technical field of fault monitoring, and relates to a data-driven intermittent process online fault monitoring technology, in particular to a batch imaging-based convolution self-encoding fault monitoring method. Background technique [0002] At present, the batch production process is developing towards refinement and intensification. It is very important to effectively monitor the whole production process, because it can not only ensure the production safety of the batch process, but also improve product quality and production efficiency, reduce energy consumption and pollute. [0003] The most commonly used methods in intermittent process fault monitoring research are multivariate statistical methods with multiway principal component analysis (MPCA) and multiway partial least squares (MPLS) as the core. By constructing T 2 (Hotelling-T 2 ) and SPE (squared prediction error) statistics, and compared with the statistical...

AI Technical Summary

Problems solved by technology

This method was initially applied to image recognition, and in recent years has been gradually extended to abnormal detection of images, videos, wafers, etc., but it is rarely used in fault monitoring of intermittent processes.

Images