Intelligent mechanical fault diagnosis using multiscale residual network and multisensor fusion

Measurement Science and Technology, Год журнала: 2024, Номер 35(11), С. 116007 - 116007

Опубликована: Авг. 1, 2024

Abstract Mechanical faults in manufacturing systems need to be diagnosed accurately ensure safety and cost savings. With the development of sensor technologies, data from multiple sensors is frequently utilized assess health intricate industrial systems. In such cases, it necessary study multisensor based intelligent mechanical fault diagnosis method. First, converted into grey images then fused a three-channel red-green-blue (RGB) image. Then, multiscale with residual convolution module proposed, which can extract deep features complex raw signal. Additionally, an attention for channel spatial introduced adaptively adjust feature response values each scale. Two datasets specific engineering application are used validate superiority network. The results show that network outperforms other networks terms identification accuracy, diagnostic efficiency, applicability.

Язык: Английский

---

Multi-scale quadratic convolutional neural network for bearing fault diagnosis based on multi-sensor data fusion

Nonlinear Dynamics, Год журнала: 2025, Номер unknown

Опубликована: Фев. 17, 2025

---

A graph representation learning-based method for fault diagnosis of rotating machinery under time-varying speed conditions

Nonlinear Dynamics, Год журнала: 2025, Номер unknown

Опубликована: Апрель 11, 2025

Язык: Английский

---

Complex Working Condition Bearing Fault Diagnosis Based on Multi-Feature Fusion and Improved Weighted Balance Distribution Adaptive Approach

Lubricants, Год журнала: 2025, Номер 13(5), С. 221 - 221

Опубликована: Май 15, 2025

In order to improve the accuracy and generalization ability of fault diagnosis for rotating machinery bearings under complex working conditions, a new model based on multi-feature fusion improved weighted balance distribution adaptation is proposed. Firstly, an optimized variational mode decomposition algorithm introduced denoise signal. Secondly, in complement information from multiple dimensions, thirteen frequency features four entropy are extracted. Then, 17 directly concatenated by dimension form high-dimensional feature vector that better adapts conditions modes. Finally, adaptive used reduce difference between source domain target domain. K-nearest neighbors as classifier determine category. Using Case Western Reserve University dataset validation, experimental results show proposed achieves average diagnostic 99.34% 12 conditions.

Язык: Английский

---

A class-center fine-tuning prototypical network for few-shot fault diagnosis of turnout switch machine driven by multi-source signals

Measurement, Год журнала: 2024, Номер 242, С. 115920 - 115920

Опубликована: Окт. 10, 2024

Язык: Английский

---

A graph representation learning-based method for fault diagnosis of rotating machinery under time-varying speed conditions

Research Square (Research Square), Год журнала: 2024, Номер unknown

Опубликована: Ноя. 29, 2024

The health of rotating machinery is critical to the quality and efficiency manufacturing process. However, existing intelligent fault diagnosis methods are mostly carried out under constant speed conditions, which makes it difficult adapt variability complexity equipment with time in actual industrial scenarios. Based on graph learning self-attention mechanism, this study proposes a novel method for time-varying conditions. Node feature information extracted from raw vibration signals multiple directions construct spatial data. Then transformed into embedded data, spatiotemporal nested containing built. After that, convolutional attention interactive parallel network model established. Combining advantages contained deeply mined promote identify types correctly. superiority proposed verified by two speeds test Compared other deep methods, can still achieve optimal diagnostic results even case insufficient training samples.

Язык: Английский

---