Intelligent Performance Degradation Prediction of Light-Duty Gas Turbine Engine Based on Limited Data

Symmetry, Journal Year: 2025, Volume and Issue: 17(2), P. 277 - 277

Published: Feb. 11, 2025

The health monitoring system has been the main technological approach to extending life of gas turbine engines and reducing maintenance costs resulting from performance degradation caused by asymmetric factors like carbon deposition, damage, or deformation. One most critical techniques within is prediction. At present, research on prediction carried out using NASA’s open dataset, C-MAPSS, without considering that measurements are not always available, as in ideal dataset. This limitation makes fault diagnosis algorithms remaining useful methods difficult apply real engines. Therefore, solve problem light-duty engines, a diagram proposed based Long Short-Term Memory (LSTM). Various types onboard signals taken into consideration among experimental data. Only accumulated usage time, total temperature pressure before inlet, low-pressure rotor speed, high-pressure fuel flow rate, exhaust temperature, thrust used training process, which indispensable for an aero-engine. A genetic algorithm (GA) introduced process optimize hyperparameters LSTM. modeled with GA-LSTM method validated maximum error 70 daN, mean absolute percentage (MAPE) less than 0.04. study provides practical implementing systems improve engine reliability, economy, environmental performance.

Language: Английский

---

Dual-graph regularized sparse robust adaptive concept factorization

Knowledge-Based Systems, Journal Year: 2025, Volume and Issue: unknown, P. 113299 - 113299

Published: March 1, 2025

---

Low-Rank and Discriminate Block Diagonal Representation via Projection Reconstruction for Image Classification

Digital Signal Processing, Journal Year: 2025, Volume and Issue: unknown, P. 105194 - 105194

Published: March 1, 2025

Language: Английский

---

A unified temporal link prediction framework based on nonnegative matrix factorization and graph regularization

The Journal of Supercomputing, Journal Year: 2025, Volume and Issue: 81(6)

Published: April 23, 2025

Language: Английский

---

Robust Face Recognition Based on the Wing Loss and the ℓ1 Penalty

Electronics, Journal Year: 2025, Volume and Issue: 14(9), P. 1736 - 1736

Published: April 24, 2025

In recent years, face recognition under occluded or corrupted conditions has emerged as a prominent research topic. The advancement in sparse sampling techniques based on regression analysis provided novel solution to this challenge. Currently, numerous regression-based models have been investigated by researchers address problem. However, the accuracy of most existing deteriorates significantly when handling heavily severely facial images. To overcome limitation, paper proposes wing-constrained coding (WCSC) model and its weighted variant (weighted coding, WWCSC) for robust complex scenarios. corresponding minimization problems are solved using alternating direction method multipliers (ADMM) algorithm. Extensive experiments conducted four benchmark databases: Olivetti Research Laboratory (ORL) database, Yale AR database Face Recognition Technology (FERET) evaluate proposed method’s performance. Comparative results demonstrate that WWCSC maintains superior rates even challenging involving significant occlusion corruption, highlighting remarkable robustness tasks. This study provides both theoretical empirical validation effectiveness approach.

Language: Английский

---

Lightweight Solution for English Course Quality Evaluation Supported by Edge Intelligence

Internet Technology Letters, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 9, 2025

ABSTRACT With the development of artificial intelligence and computer vision technologies, smart education supported by has gradually replaced traditional classroom teaching mode, such as English courses. However, existing studies about mainly focus on student behavior recognition neglect quality evaluation learning in classroom. On other hand, it still lacks a lightweight solution for network environment with limited resources. This paper establishes course scheme to tackle above issues. First, camera is used collect information class; then, network, which consists MobileNet v4 Small YOLO deployed edge computing device automatically process from attendance rate estimation students' recognition; lastly, server remote control room collects all classrooms implement evaluation. The experimental simulated results show that detection can meet requirements real application.

Language: Английский

---

Discriminative Approximate Regression Projection for Feature Extraction

Information Fusion, Journal Year: 2025, Volume and Issue: unknown, P. 103088 - 103088

Published: March 1, 2025

Language: Английский

---