DCGAN-based linear contour process quality abnormity monitoring method and system

Abstract

The invention relates to a DCGAN-based linear contour process quality abnormity monitoring method and system, and the method comprises an offline training stage and an online monitoring stage, in the offline training stage, linear contour process data in a historical normal operation state and an abnormal state are respectively collected, samples are amplified, and the linear contour process data in the abnormal state are acquired; in the online monitoring stage, the convolutional neural network is trained based on the amplified samples, process anomaly detection is performed according to the trained convolutional neural network, and the problem of limited manual feature extraction can be solved by applying the deep convolutional generative adversarial network to the monitoring of linear contour process quality anomaly. According to the invention, the method can improve the monitoring efficiency of the quality abnormality of the linear contour process, can train the historical data under the condition of an unbalanced data set, guarantees the recognition precision of an abnormal process with a small sample data size, and can also effectively improve the monitoring efficiency of the quality abnormality of the linear contour process through employing the convolutional neural network.

Description

technical field [0001] The invention relates to a DCGAN-based method and system for monitoring abnormality of linear profile process quality. Background technique [0002] Sensors used to measure equipment parameters and quality characteristics such as displacement, pressure, temperature, humidity, and size are installed in large numbers in modern production processes, and the collected process data gradually develops from a single value in the past to complex time series or space series data. When the data representing the operating state of the process can be approximated by a linear model, it is called a linear profile. How to discover the internal operation rules of the system from a large amount of linear profile data and build a real-time and intelligent quality monitoring model suitable for modern production processes has become a key issue to ensure safe production and reduce quality costs. [0003] In the research related to linear profile monitoring, monitoring me...

AI Technical Summary

Problems solved by technology

However, the current research on contour monitoring based on intelligent methods all use shallow models that do not exceed two layers of nonlinear feature transformation, and there are still problems such as limited manual feature extraction, insufficient classification accuracy and generalization ability.

Images