Fault Diagnosis for Reducers Based on a Digital Twin

Area of Science:

• Mechanical Engineering
• Computer Science
• Industrial IoT

Background:

• Existing fault diagnosis models struggle with single fault types, low similarity, and poor visualization.
• There is a need for advanced fault diagnosis methods that offer higher accuracy and better state monitoring.
• Digital twin technology presents a promising avenue for overcoming these limitations.

Purpose of the Study:

• To propose a new digital twin-based method for fault diagnosis.
• To address the shortcomings of existing fault diagnosis modeling techniques.
• To improve the accuracy and visual representation of state monitoring in fault diagnosis.

Main Methods:

• Established a fault diagnosis test platform for analyzing faults under constant and variable speeds.
• Integrated fault data into the Unity3D platform for online diagnosis and real-time updates.
• Conducted industrial tests of the digital twin model and compared its performance with other advanced methods.

Main Results:

• The proposed digital twin method achieved an overall accuracy of 99.5% for the entire reducer.
• This accuracy significantly surpasses methods focusing on individual components (e.g., bearings at 93.5%, gear shafts at 96.3%, shells at 92.6%).
• The digital twin model demonstrated high accuracy and application feasibility in industrial settings.

Conclusions:

• The digital twin-based fault diagnosis method offers superior accuracy and effectiveness compared to traditional approaches.
• This technology provides a robust solution for online, real-time fault diagnosis and state monitoring.
• The proposed method is highly feasible for industrial applications, enhancing machinery reliability.