Safety Poka Yoke in Zero-Defect Manufacturing Based on Digital Twins

In this article, the proposed work aims to further optimize the fault diagnosis effect of manufacturing equipment, explore the application of digital twins technology in intelligent manufacturing. The equipment failures in Poka Yoke technology are adopted, and a fault identification and chopping algorithm is designed based on the active learning-deep neural network (AL-DNN) and domain adversarial neural networks (DANN). In addition, a digital twins workshop management and control system is designed for intelligent manufacturing management. The experimental exploration reveals that the accuracy of the AL-DNN algorithm is as high as 99.248%, which is more in line with practical applications. The DANN algorithm can realize fault identification and diagnosis under different working conditions. Compared with other deep learning algorithms, the accuracy of the DANN can be increased by up to 20.256%, showing higher accuracy in contrast to the traditional algorithm, so the effect is more stable. In addition, the digital twins manufacturing management system designed shows good performances, which can intuitively display the specific conditions of workshop and realize basic operating functions. The concept of digital twins is innovatively introduced into equipment fault diagnosis and trend prediction, which can provide scientific and effective reference data for subsequent research on intelligent manufacturing.

Automated summary

This article aims to optimize the fault diagnosis effect of manufacturing equipment and explore the application of digital twins technology in intelligent manufacturing. By adopting equipment failures in Poka Yoke technology, a fault identification and chopping algorithm based on active learning-deep neural network (AL-DNN) and domain adversarial neural networks (DANN) is designed. Additionally, a digital twins workshop management and control system is designed for intelligent manufacturing management. Experimental exploration shows that the accuracy of the AL-DNN algorithm is as high as 99.248% and the DANN algorithm can realize fault identification and diagnosis under different working conditions, with an accuracy improvement of up to 20.256% compared to other deep learning algorithms.