A Particle Swarm Optimization-Based Ensemble Broad Learning System for Intelligent Fault Diagnosis in Safety-Critical Energy Systems with High-Dimensional Small Samples

Mathematics, Год журнала: 2025, Номер 13(5), С. 797 - 797

Опубликована: Фев. 27, 2025

Intelligent fault diagnosis (IFD) plays a crucial role in reducing maintenance costs and enhancing the reliability of safety-critical energy systems (SCESs). In recent years, deep learning-based IFD methods have achieved high accuracy extracting implicit higher-order correlations between features. However, excessive long training time learning models conflicts with requirements real-time analysis for IFD, hindering their further application practical industrial environments. To address aforementioned challenge, this paper proposes an innovative method SCES that combines particle swarm optimization (PSO) algorithm ensemble broad system (EBLS). Specifically, (BLS), known its low complexity classification accuracy, is adopted as alternative to SCES. Furthermore, EBLS designed enhance model stability high-dimensional small samples by incorporating random forest (RF) strategy into traditional BLS framework. order reduce computational cost EBLS, which constrained selection hyperparameters, PSO employed optimize hyperparameters EBLS. Finally, validated through simulated data from complex nuclear power plant (NPP). Numerical experiments reveal proposed significantly improved diagnostic efficiency while maintaining accuracy. summary, approach shows great promise boosting capabilities

Язык: Английский

---

A cross domain processing deep transfer learning network for rotating machinery fault diagnosis

Measurement Science and Technology, Год журнала: 2025, Номер 36(4), С. 046132 - 046132

Опубликована: Апрель 8, 2025

Abstract In the field of intelligent fault diagnosis mechanical equipment, existing cross-domain diagnostic models based on transfer learning (TL) do not utilise commonality information between two domains in data processing stage, which leads to loss transferable features that are essential for task. To address this issue, paper proposes a deep TL network model (CDPDTLN), consists (CDP) module, feature extraction module and domain-adaptive module. CDP adaptive multivariate variational modal decomposition algorithm is used process source target domain simultaneously, preserving common domains. realise work under various complex operating conditions, an improved multi-scale residual proposed extract domain-invariant features. combined distribution adaptation (CDDA) strategy align marginal conditional distributions CDDA strategy, weighted mean square discrepancy metric defined by combining maximum with enhance alignment confusion capabilities. multi-scenario experiments, accuracy CDPDTLN exceeds 95%. The results show can effectively retain learn features, significantly improving reliability robustness diagnosis.

Язык: Английский

---

Advanced fault detection, diagnosis and prognosis in HVAC systems: Lifecycle insight, key challenges, and promising approaches

Renewable and Sustainable Energy Reviews, Год журнала: 2025, Номер 219, С. 115867 - 115867

Опубликована: Май 27, 2025

Язык: Английский

---

Novel Machine Learning Paradigms-Enabled Methods for Smart Building Operations in Data-Challenging Contexts: Progress and Perspectives

National Science Open, Год журнала: 2024, Номер 3(3), С. 20230068 - 20230068

Опубликована: Фев. 2, 2024

Язык: Английский

---