A review of artificial intelligence application for machining surface quality prediction: from key factors to model development

Journal of Intelligent Manufacturing, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 10, 2025

Language: Английский

---

Uncertainty-Aware Self-Attention Model for Time Series Prediction with Missing Values

Fractal and Fractional, Journal Year: 2025, Volume and Issue: 9(3), P. 181 - 181

Published: March 16, 2025

Missing values in time series data present a significant challenge, often degrading the performance of downstream tasks such as classification and forecasting. Traditional approaches address this issue by first imputing missing then independently solving predictive tasks. Recent methods have leveraged self-attention models to enhance imputation quality accelerate inference. These models, however, predict based on all input observations—including values—thereby potentially compromising fidelity imputed data. In paper, we propose Uncertainty-Aware Self-Attention (UASA) model overcome these limitations. Our approach introduces two novel techniques: (i) A mechanism with partially observed diagonal that effectively captures complex non-local dependencies data—a characteristic also fractional-order systems. This draws inspiration from fractional calculus, where non-integer-order derivatives better characterize dynamical systems long-memory effects, providing more comprehensive mathematical framework for handling temporal And (ii) uncertainty quantification inform The UASA comprises an upstream component prediction, trained jointly end-to-end fashion optimize both accuracy task-specific objectives simultaneously. For tasks, demonstrates remarkable even under high rates, achieving ROC-AUC 99.5%, PR-AUC 58.5%, F1-SCORE 49.3%. forecasting AUST-Gait dataset, achieves Mean Squared Error (MSE) 0.72 0% conditions (i.e., complete input). Under training strategy evaluated across average MSE 0.74, showcasing its adaptability robustness diverse scenarios.

Language: Английский

---

An explainable multi-layer graph attention network for product completion time prediction in aircraft final assembly lines

Journal of Manufacturing Systems, Journal Year: 2025, Volume and Issue: 80, P. 1053 - 1071

Published: May 13, 2025

Language: Английский

---

A systematic quality-integrated diagnostic method for complex product assembly using multi-task spatial–temporal transfer learning

The International Journal of Advanced Manufacturing Technology, Journal Year: 2024, Volume and Issue: 135(3-4), P. 1355 - 1375

Published: Oct. 8, 2024

Abstract The assembly process is generally considered one of the primary factors influencing quality complex products. Currently, most existing quality-integrated diagnostic methods for products tend to deteriorate over different processes and degrade time. To address this issue, paper introduces a systematic method product processes. First, influence error sources in are analyzed using 5M1E FAHP methodologies. Next, similarity-based multi-task clustering dismantling RGMM applied divide tasks. Finally, MMD-MSE employed develop prediction model spatial–temporal transfer learning approach. Experiments were conducted on an array antenna task, comparing proposed with conventional methods. results show that accuracy PrUP 97.6% 95.2%, respectively, processes, fluctuation less than 6%. effectively meet expert evaluations provide stable, reliable, practical solution addressing fluctuations production

Language: Английский

---

A human-inspired slow-fast dual-branch method for product quality prediction of complex manufacturing processes with hierarchical variations

Advanced Engineering Informatics, Journal Year: 2024, Volume and Issue: 64, P. 102967 - 102967

Published: Dec. 4, 2024

Language: Английский

---