A Weighted Collaborative Prediction Method of RUL for Integrated Circuits in Multi-Failure Modes Based on Transformer

• Published in: IFAC-PapersOnLine 2023-01, Vol.56 (2), p.7632-7637
• Main Authors: Zhang, Zibo, Jia, Long, Zou, Xinyu, Chen, Chang, Chen, Wei, Lan, Chunyun, Ma, Jian
• Format: Article
• Language: English
• Subjects: Integrated circuits; locally weighted regression method; multi-failure modes; RUL prediction; Transformer deep neural network; weighted collaborative prediction
• ISSN: 2405-8963; 2405-8963
• DOI: 10.1016/j.ifacol.2023.10.1161

Accurately predicting the Remaining Useful Life (RUL) of integrated circuits in multi-failure modes is critical for ensuring the safe and efficient operation of electronic devices. In this paper, we propose a novel RUL prediction method that combines a locally weighted regression method, a Transformer deep neural network, and a multi-failure modes weighted collaborative prediction approach. We first use a locally weighted regression method to eliminate noise from the original feature time series. Then, the feature time series is input into the proposed classification model to classify the failure modes and establish the failure mode weighted function. Finally, the feature time series is fed into the proposed regression model for RUL prediction with weighted processing to obtain accurate RUL predictions. Our experimental results demonstrate that our method effectively captures the degradation information of integrated circuits and generates precise RUL predictions. We demonstrate the efficacy of our method on a circuit stimulation model, showing that it outperforms several state-of-the-art RUL prediction methods. Our proposed method has the potential to enhance the reliability and safety of electronic devices in various applications.