International Journal of Adaptive Control and Signal Processing, Год журнала: 2024, Номер unknown

Опубликована: Сен. 15, 2024

ABSTRACT With the rapid development of industrial technology, processes become increasingly complex, presenting characteristics large‐scale and multi‐unit collaboration. However, most current fault detection methods focus on nonlinearity, dynamics, other characteristics, while neglecting spatiotemporal information. To address this issue, an adaptive decouple graph convolutional network based quality‐related method is proposed in article. First, temporal spatial are combined organically form joint training. Second, considering that fixed structures cannot reflect dynamic relationships among nodes, we weighted mask mechanism to construct correlation embedded with priori knowledge. Multi‐attention used integrate information, besides, designed a decoupling layer avoid information redundancy. Finally, establish regression model, latent variables extracted by layer, statistic constructed Kullback–Leibler divergence. The effectiveness feasibility illustrated hot strip mill process Tennessee Eastman process.

Язык: Английский